system/freebsd-src
system/freebsd-src
1
0
Fork 0
mirror of https://github.com/freebsd/freebsd-src synced 2024-10-15 12:54:27 +00:00
Code Issues Packages Projects Releases Wiki Activity
freebsd-src/lib/libkvm/kvm_minidump_amd64.c

309 lines
8.5 KiB
C
Raw Normal View History

Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
/*-
* Copyright (c) 2006 Peter Wemm
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
Remove trailing whitespace.
2015-11-26 19:42:10 +00:00
* AMD64 machine dependent routines for kvm and minidumps.
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
*/
#include <sys/param.h>
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
#include <sys/endian.h>
#include <stdint.h>
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
#include <stdlib.h>
Fix missing includes of <string.h>, to silence some compiler warnings. Submitted by: Pawel Worach
2009-06-14 12:42:06 +00:00
#include <string.h>
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
#include <unistd.h>
#include <kvm.h>
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
#include "../../sys/amd64/include/minidump.h"
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
#include <limits.h>
#include "kvm_private.h"
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
#include "kvm_amd64.h"
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
#define amd64_round_page(x) roundup2((kvaddr_t)(x), AMD64_PAGE_SIZE)
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
struct vmstate {
struct minidumphdr hdr;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
amd64_pte_t *page_map;
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
};
static int
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_amd64_minidump_probe(kvm_t *kd)
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
{
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (_kvm_probe_elf_kernel(kd, ELFCLASS64, EM_X86_64) &&
_kvm_is_minidump(kd));
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
static void
_amd64_minidump_freevtop(kvm_t *kd)
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
{
struct vmstate *vm = kd->vmst;
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
free(vm->page_map);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
free(vm);
kd->vmst = NULL;
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
static int
_amd64_minidump_initvtop(kvm_t *kd)
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
{
struct vmstate *vmst;
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
off_t off, sparse_off;
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
vmst = _kvm_malloc(kd, sizeof(*vmst));
Fix up pointer issues with lib/libkvm In particular, - avoid dereferencing NULL pointers - test pointers against NULL, not 0 - test for errout == NULL in the top-level functions (kvm_open, kvm_openfiles, kvm_open2, etc) - Replace a realloc and free on failure with reallocf Found with: devel/cocchinelle Differential Revision: https://reviews.freebsd.org/D5954 MFC after: 1 week Reviewed by: jhb Sponsored by: EMC / Isilon Storage Division
2016-04-22 18:05:34 +00:00
if (vmst == NULL) {
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
_kvm_err(kd, kd->program, "cannot allocate vm");
return (-1);
}
kd->vmst = vmst;
if (pread(kd->pmfd, &vmst->hdr, sizeof(vmst->hdr), 0) !=
sizeof(vmst->hdr)) {
_kvm_err(kd, kd->program, "cannot read dump header");
return (-1);
}
if (strncmp(MINIDUMP_MAGIC, vmst->hdr.magic, sizeof(vmst->hdr.magic)) != 0) {
_kvm_err(kd, kd->program, "not a minidump for this platform");
return (-1);
}
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
/*
* NB: amd64 minidump header is binary compatible between version 1
* and version 2; this may not be the case for the future versions.
*/
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
vmst->hdr.version = le32toh(vmst->hdr.version);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (vmst->hdr.version != MINIDUMP_VERSION && vmst->hdr.version != 1) {
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
_kvm_err(kd, kd->program, "wrong minidump version. expected %d got %d",
MINIDUMP_VERSION, vmst->hdr.version);
return (-1);
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
vmst->hdr.msgbufsize = le32toh(vmst->hdr.msgbufsize);
vmst->hdr.bitmapsize = le32toh(vmst->hdr.bitmapsize);
vmst->hdr.pmapsize = le32toh(vmst->hdr.pmapsize);
vmst->hdr.kernbase = le64toh(vmst->hdr.kernbase);
vmst->hdr.dmapbase = le64toh(vmst->hdr.dmapbase);
vmst->hdr.dmapend = le64toh(vmst->hdr.dmapend);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
/* Skip header and msgbuf */
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
off = AMD64_PAGE_SIZE + amd64_round_page(vmst->hdr.msgbufsize);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
sparse_off = off + amd64_round_page(vmst->hdr.bitmapsize) +
amd64_round_page(vmst->hdr.pmapsize);
if (_kvm_pt_init(kd, vmst->hdr.bitmapsize, off, sparse_off,
AMD64_PAGE_SIZE, sizeof(uint64_t)) == -1) {
_kvm_err(kd, kd->program, "cannot load core bitmap");
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
return (-1);
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
off += amd64_round_page(vmst->hdr.bitmapsize);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
vmst->page_map = _kvm_malloc(kd, vmst->hdr.pmapsize);
if (vmst->page_map == NULL) {
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
_kvm_err(kd, kd->program, "cannot allocate %d bytes for page_map",
vmst->hdr.pmapsize);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
return (-1);
}
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (pread(kd->pmfd, vmst->page_map, vmst->hdr.pmapsize, off) !=
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
(ssize_t)vmst->hdr.pmapsize) {
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
_kvm_err(kd, kd->program, "cannot read %d bytes for page_map",
vmst->hdr.pmapsize);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
return (-1);
}
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
off += amd64_round_page(vmst->hdr.pmapsize);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
return (0);
}
static int
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_amd64_minidump_vatop_v1(kvm_t *kd, kvaddr_t va, off_t *pa)
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
{
struct vmstate *vm;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
amd64_physaddr_t offset;
amd64_pte_t pte;
kvaddr_t pteindex;
amd64_physaddr_t a;
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
off_t ofs;
vm = kd->vmst;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
offset = va & AMD64_PAGE_MASK;
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
if (va >= vm->hdr.kernbase) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
pteindex = (va - vm->hdr.kernbase) >> AMD64_PAGE_SHIFT;
Add bounds check for pteindex / pdeindex
2014-02-12 15:57:38 +00:00
if (pteindex >= vm->hdr.pmapsize / sizeof(*vm->page_map))
goto invalid;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
pte = le64toh(vm->page_map[pteindex]);
if ((pte & AMD64_PG_V) == 0) {
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop_v1: pte not valid");
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
goto invalid;
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
a = pte & AMD64_PG_FRAME;
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
ofs = _kvm_pt_find(kd, a);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
if (ofs == -1) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop_v1: physical address 0x%jx not in minidump",
(uintmax_t)a);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
goto invalid;
}
*pa = ofs + offset;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (AMD64_PAGE_SIZE - offset);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
} else if (va >= vm->hdr.dmapbase && va < vm->hdr.dmapend) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
a = (va - vm->hdr.dmapbase) & ~AMD64_PAGE_MASK;
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
ofs = _kvm_pt_find(kd, a);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
if (ofs == -1) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop_v1: direct map address 0x%jx not in minidump",
(uintmax_t)va);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
goto invalid;
}
*pa = ofs + offset;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (AMD64_PAGE_SIZE - offset);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
} else {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop_v1: virtual address 0x%jx not minidumped",
(uintmax_t)va);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
goto invalid;
}
invalid:
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, 0, "invalid address (0x%jx)", (uintmax_t)va);
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
return (0);
}
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
static int
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_amd64_minidump_vatop(kvm_t *kd, kvaddr_t va, off_t *pa)
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
{
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
amd64_pte_t pt[AMD64_NPTEPG];
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
struct vmstate *vm;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
amd64_physaddr_t offset;
amd64_pde_t pde;
amd64_pte_t pte;
kvaddr_t pteindex;
kvaddr_t pdeindex;
amd64_physaddr_t a;
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
off_t ofs;
vm = kd->vmst;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
offset = va & AMD64_PAGE_MASK;
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (va >= vm->hdr.kernbase) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
pdeindex = (va - vm->hdr.kernbase) >> AMD64_PDRSHIFT;
Add bounds check for pteindex / pdeindex
2014-02-12 15:57:38 +00:00
if (pdeindex >= vm->hdr.pmapsize / sizeof(*vm->page_map))
goto invalid;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
pde = le64toh(vm->page_map[pdeindex]);
if ((pde & AMD64_PG_V) == 0) {
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop: pde not valid");
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
goto invalid;
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
if ((pde & AMD64_PG_PS) == 0) {
a = pde & AMD64_PG_FRAME;
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
/* TODO: Just read the single PTE */
ofs = _kvm_pt_find(kd, a);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (ofs == -1) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
"cannot find page table entry for %ju",
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
(uintmax_t)a);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
goto invalid;
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
if (pread(kd->pmfd, &pt, AMD64_PAGE_SIZE, ofs) !=
AMD64_PAGE_SIZE) {
_kvm_err(kd, kd->program,
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
"cannot read page table entry for %ju",
(uintmax_t)a);
goto invalid;
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
pteindex = (va >> AMD64_PAGE_SHIFT) &
(AMD64_NPTEPG - 1);
pte = le64toh(pt[pteindex]);
if ((pte & AMD64_PG_V) == 0) {
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop: pte not valid");
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
goto invalid;
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
a = pte & AMD64_PG_FRAME;
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
} else {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
a = pde & AMD64_PG_PS_FRAME;
a += (va & AMD64_PDRMASK) ^ offset;
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
}
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
ofs = _kvm_pt_find(kd, a);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (ofs == -1) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop: physical address 0x%jx not in minidump",
(uintmax_t)a);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
goto invalid;
}
*pa = ofs + offset;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (AMD64_PAGE_SIZE - offset);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
} else if (va >= vm->hdr.dmapbase && va < vm->hdr.dmapend) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
a = (va - vm->hdr.dmapbase) & ~AMD64_PAGE_MASK;
libkvm: Improve physical address lookup scaling. Instead of using a hash table to convert physical page addresses to offsets in the sparse page array, cache the number of bits set for each 4MB chunk of physical pages. Upon lookup, find the nearest cached population count, then add/subtract the number of bits from that point to the page's PTE bit. Then multiply by page size and add to the sparse page map's base offset. This replaces O(n) worst-case lookup with O(1) (plus a small number of bits to scan in the bitmap). Also, for a 128GB system, a typical kernel core of about 8GB will now only require ~4.5MB of RAM for this approach instead of ~48MB as with the hash table. More concretely, /usr/sbin/crashinfo against the same core improves from a max RSS of 188MB and wall time of 43.72s (33.25 user 2.94 sys) to 135MB and 9.43s (2.58 user 1.47 sys). Running "thread apply all bt" in kgdb has a similar RSS improvement, and wall time drops from 4.44s to 1.93s. Reviewed by: jhb Sponsored by: Backtrace I/O
2016-07-18 01:55:25 +00:00
ofs = _kvm_pt_find(kd, a);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (ofs == -1) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop: direct map address 0x%jx not in minidump",
(uintmax_t)va);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
goto invalid;
}
*pa = ofs + offset;
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (AMD64_PAGE_SIZE - offset);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
} else {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, kd->program,
"_amd64_minidump_vatop: virtual address 0x%jx not minidumped",
(uintmax_t)va);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
goto invalid;
}
invalid:
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, 0, "invalid address (0x%jx)", (uintmax_t)va);
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
return (0);
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
static int
_amd64_minidump_kvatop(kvm_t *kd, kvaddr_t va, off_t *pa)
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
{
if (ISALIVE(kd)) {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
_kvm_err(kd, 0,
"_amd64_minidump_kvatop called in live kernel!");
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
return (0);
}
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
if (((struct vmstate *)kd->vmst)->hdr.version == 1)
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (_amd64_minidump_vatop_v1(kd, va, pa));
amd64: introduce minidump version 2 After KVA space was increased to 512GB on amd64 it became impractical to use PTEs as entries in the minidump map of dumped pages, because size of that map alone would already be 1GB. Instead, we now use PDEs as page map entries and employ two stage lookup in libkvm: virtual address -> PDE -> PTE -> physical address. PTEs are now dumped as regular pages. Fixed page map size now is 2MB. libkvm keeps support for accessing amd64 minidumps of version 1. Support for 1GB pages is added. Many thanks to Alan Cox for his guidance, numerous reviews, suggestions, enhancments and corrections. Reviewed by: alc [kernel part] MFC after: 15 days
2010-11-11 18:35:28 +00:00
else
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
return (_amd64_minidump_vatop(kd, va, pa));
Teach libkvm how to read minidumps. It turns out that minidumps are far more convenient for libkvm to work with because of the page table block at the beginning. As a result, the MD code is smaller. libkvm will automatically detect old vs mini dumps on i386 and amd64. libkvm will handle i386 PAE and non-PAE modes. There is a PAE flag in the i386 minidump header to signal the width of the entries in the page table block. Other convenient values are also present, such as kernbase and the direct map addresses on amd64.
2006-04-21 04:32:51 +00:00
}
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
libkvm: fix warning issues post-r291406 - Fix -Wunused warnings with *_native detection handlers by marking `kd` __unused, except with arm/mips, where a slightly more complicated scheme is required to handle the native case vs the non-native case. - Fix -Wmissing-variable-declarations warnings by marking struct kvm_arch objects static. Differential Revision: D10071 MFC after: 1 week Reviewed by: vangyzen Tested with: WIP test code (D10024) // kgdb7121 (i386 crash/kernel on amd64) Sponsored by: Dell EMC Isilon
2017-03-22 02:42:00 +00:00
static struct kvm_arch kvm_amd64_minidump = {
Add support to libkvm for reading vmcores from other architectures. - Add a kvaddr_type to represent kernel virtual addresses instead of unsigned long. - Add a struct kvm_nlist which is a stripped down version of struct nlist that uses kvaddr_t for n_value. - Add a kvm_native() routine that returns true if an open kvm descriptor is for a native kernel and memory image. - Add a kvm_open2() function similar to kvm_openfiles(). It drops the unused 'swapfile' argument and adds a new function pointer argument for a symbol resolving function. Native kernels still use _fdnlist() from libc to resolve symbols if a resolver function is not supplied, but cross kernels require a resolver. - Add a kvm_nlist2() function similar to kvm_nlist() except that it uses struct kvm_nlist instead of struct nlist. - Add a kvm_read2() function similar to kvm_read() except that it uses kvaddr_t instead of unsigned long for the kernel virtual address. - Add a new kvm_arch switch of routines needed by a vmcore backend. Each backend is responsible for implementing kvm_read2() for a given vmcore format. - Use libelf to read headers from ELF kernels and cores (except for powerpc cores). - Add internal helper routines for the common page offset hash table used by the minidump backends. - Port all of the existing kvm backends to implement a kvm_arch switch and to be cross-friendly by using private constants instead of ones that vary by platform (e.g. PAGE_SIZE). Static assertions are present when a given backend is compiled natively to ensure the private constants match the real ones. - Enable all of the existing vmcore backends on all platforms. This means that libkvm on any platform should be able to perform KVA translation and read data from a vmcore of any platform. Tested on: amd64, i386, sparc64 (marius) Differential Revision: https://reviews.freebsd.org/D3341
2015-11-27 18:58:26 +00:00
.ka_probe = _amd64_minidump_probe,
.ka_initvtop = _amd64_minidump_initvtop,
.ka_freevtop = _amd64_minidump_freevtop,
.ka_kvatop = _amd64_minidump_kvatop,
.ka_native = _amd64_native,
};
KVM_ARCH(kvm_amd64_minidump);
Reference in a new issue Copy permalink