This commit refactors the UMA small alloc code and
removes most UMA machine-dependent code.
The existing machine-dependent uma_small_alloc code is almost identical
across all architectures, except for powerpc where using the direct
map addresses involved extra steps in some cases.
The MI/MD split was replaced by a default uma_small_alloc
implementation that can be overridden by architecture-specific code by
defining the UMA_MD_SMALL_ALLOC symbol. Furthermore, UMA_USE_DMAP was
introduced to replace most UMA_MD_SMALL_ALLOC uses.
Reviewed by: markj, kib
Approved by: markj (mentor)
Differential Revision: https://reviews.freebsd.org/D45084
Implement everything currently in vm_radix.c with calls to functions
in subr_pctrie.c, asccessed via the interface provided by the
DEFINE_PCTRIE_SMR macro.
Add back some #includes removed in the first attempt, and avoid the
use of a discontinued type in a bit of conditionally compiled code.
Reviewed by: alc, markj
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D41344
Implement everything currently in vm_radix.c with calls to functions
in subr_pctrie.c, asccessed via the interface provided by the
DEFINE_PCTRIE_SMR macro.
Reviewed by: alc, markj
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D41344
The computation of keybarr(), the function that determines when a
search has failed at a non-leaf node, can be done in a way that
computes the 'slot' value when keybarr() fails, which is exactly when
slot() would next be invoked. Computing things this way saves space in
search loops.
This reduces the amd64 coding of the search loop in vm_radix_lookup
from 40 bytes to 28 bytes.
Reviewed by: alc
Tested by: pho (as part of a larger change)
Differential Revision: https://reviews.freebsd.org/D41235
The clev field in the node struct is almost always multiplied by
WIDTH; occasionally, it is incremented and then multiplied by
WIDTH. Instructions can be saved by storing it always multiplied by
WIDTH.
For the computation of slot(), this just eliminates a
multiplication. For trimkey(), where the caller always adds one to
clev before passing it as an argument, this change has the caller, not
the caller, do that. Trimkey() handles it not by adding WIDTH to the
input parameter, but by shifting COUNT, and not 1. That produces the
same result, and it relieves keybarr of the need to test to avoid
shifting by more than 63 bits, since level is always <= 63.
This takes 3 instrutions and 14 bytes out of the basic lookup loop on
amd64.
Reviewed by: kib
Tested by: pho (as part of a larger change)
Differential Revision: https://reviews.freebsd.org/D41226
NULL (non-leaf) pointers with NULL leaves, there is a NULL test
removed from every iteration of an index-based search loop.
This speeds up radix trie searches by few percent. If there are any
radix tries that are not initialized with the init() function, but
instead depend on zeroing everything being proper initialization, this
will break those tries.
Reviewed by: alc, kib
Tested by: pho (as part of a larger change)
Differential Revision: https://reviews.freebsd.org/D41171
Replace the implementations of lookup_le and lookup_ge with ones
that do not use a stack or climb back up the tree, and instead
exploit the popmap field to quickly identify the place to resume
searching if the straightforward indexed search fails.
The code size of the original functions shrinks by a combined 160
bytes on amd64, and the cumulative cycle count per invocation of
the two functions together is reduced 20% in a buildworld test.
Reviewed by: alc, markj
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D40936
Two cases in the insert routine are written differently, when
they're really doing the same thing. Writing that case only once
saves 208 bytes in the compiled vm_radix_insert code and reduces
instructions executed by about 2%.
Reviewed by: alc
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D40807
Replace the 'count' field in a trie node with a bitmap that
identifies non-NULL children. Drop the 'last' field, and use the
last bit set in the bitmap instead. In lookup_le, lookup_ge,
remove, and reclaim_all, use the bitmap to find the
previous/next/only/every non-null child in constant time by
examining the bitmask instead of looping across array elements
and null-checking them one-by-one.
A buildworld test suggests that this reduces the cycle count on
those functions that eliminate some null-checks by 4.9%, 1.5%,
0.0% and 13.3%.
Reviewed by: alc
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D40775
Let node_get calculate it's own owner value. Don't pass the count
parameter, since it's always 2. Save 16 bytes in insert(). Move,
without modifying, slot and trimkey to handle use-before-declaration
problem.
Reviewed by: markj
Differential Revision: https://reviews.freebsd.org/D40723
This is purely a cosmetic change. vm_radix.c has lines that reach past
column 80 and this change cleans that up. The associated changes to
subr_pctrie.c are just to keep mirroring vm_radix.c.
Reviewed by: markj
Differential Revision: https://reviews.freebsd.org/D40764
In _lookup_ge, where a loop "looks for an available edge or val within
the current bisection node" (to quote the code comment), the value of
index has already been modified to guarantee that it is the least
value than can be found in the non-NULL child node being
examined. Therefore, if the non-NULL child is a leaf, there's no need
to compare 'index' to anything, and the value can just be returned.
The same is true for _lookup_le with 'most' replacing 'least'.
Reviewed by: alc
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D40746
Replacing a branch and two shifts with a single masking operation saves 64 bytes the pair of functions lookup_le and lookup_ge on amd64. Refresh the associated comments.
Reviewed by: alc
Differential Revision: https://reviews.freebsd.org/D40722
In the vm_radix:remove loop that searches for the last child, load
that child once, without loading it again after the search is over.
Change KASSERTS from index check to NULL node check.
Reviewed by: alc
Differential Revision: https://reviews.freebsd.org/D40721
Replace boolean_t with bool in vm_radix.c. Drop the unused function
vm_radix_is_singleton, which is unused and has no corresponding
function in subr_pctrie.c.
Reviewed by: alc
Differential Revision: <https://reviews.freebsd.org/D40586>
Use flsll(), instead of a loop, to find where two keys differ, and
then arithmetic to transform that to a trie level.
Approved by: alc, markj
Differential Revision: https://reviews.freebsd.org/D40585
The SPDX folks have obsoleted the BSD-2-Clause-FreeBSD identifier. Catch
up to that fact and revert to their recommended match of BSD-2-Clause.
Discussed with: pfg
MFC After: 3 days
Sponsored by: Netflix
__builtin_unreachable doesn't raise any compile-time warnings/errors on its
own, so problems with its usage can't be easily detected. While it would be
nice for this situation to change and compilers to at least add a warning
for trivial cases where local state means the instruction can't be reached,
this isn't the case at the moment and likely will not happen.
This commit adds an __assert_unreachable, whose intent is incredibly clear:
it asserts that this instruction is unreachable. On INVARIANTS builds, it's
a panic(), and on non-INVARIANTS it expands to __unreachable().
Existing users of __unreachable() are converted to __assert_unreachable,
to improve debuggability if this assumption is violated.
Reviewed by: mjg
Differential Revision: https://reviews.freebsd.org/D23793
The intent is to provide a header that can be included by other headers
without introducing too much pollution. smr.h depends on various
headers and will likely grow over time, but is less likely to be
required by system headers.
Rename SMR_TYPE_DECLARE() to SMR_POINTER():
- One might use SMR to protect more than just pointers; it
could be used for resizeable arrays, for example, so TYPE seems too
generic.
- It is useful to be able to define anonymous SMR-protected pointer
types and the _DECLARE suffix makes that look wrong.
Reviewed by: jeff, mjg, rlibby
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D23988
This provides the needed hint to GCC and offers an annotation for readers to
observe that it's in-fact impossible to hit this point. We'll get hit with a
a -Wswitch error if the enum applicable to the switch above were to get
expanded without the new value(s) being handled.
possible enum in a switch statement. I verified that this emits nothing
as expected on clang. radix relies on constant propagation to eliminate
any branching from these access routines.
Reported by: lwhsu/tinderbox
The tree is kept correct for readers with store barriers and careful
ordering. The existing object lock serializes writers. Consumers
will be introduced in later commits.
Reviewed by: markj, kib
Differential Revision: https://reviews.freebsd.org/D23446
Allocation explicitely initialized the 3 leading fields. The rest is an
array which is supposed to be NULL-ed prior to deallocation.
Delegate zeroing to the infrequently called object initializator.
This gets rid of one of the most common memset consumers.
Reviewed by: markj
Differential Revision: https://reviews.freebsd.org/D15989
o Call uma_startup1() after initializing kmem, vmem and domains.
o Include 8 eight VM startup pages into uma_startup_count() calculation.
o Account for vmem_startup() and vm_map_startup() preallocating pages.
o Account for extra two allocations done by kmem_init() and vmem_create().
o Hardcode the place of execution of vm_radix_reserve_kva(). Using SYSINIT
allowed several other SYSINITs to sneak in before it, thus bumping
requirement for amount of boot pages.
Mainly focus on files that use BSD 2-Clause license, however the tool I
was using misidentified many licenses so this was mostly a manual - error
prone - task.
The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
No functional change intended.
similar to the kernel memory allocator.
This simplifies NUMA allocation because the domain will be known at wait
time and races between failure and sleeping are eliminated. This also
reduces boilerplate code and simplifies callers.
A wait primitive is supplied for uma zones for similar reasons. This
eliminates some non-specific VM_WAIT calls in favor of more explicit
sleeps that may be satisfied without new pages.
Reviewed by: alc, kib, markj
Tested by: pho
Sponsored by: Netflix, Dell/EMC Isilon
vm_radix trie.
Existing vm_radix_init() function is renamed to vm_radix_zinit().
Inlines moved out of the _ headers.
Reviewed by: alc, markj (previous version)
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D11661
contain a vm_page_t at the specified index. However, with this
change, vm_radix_remove() no longer panics. Instead, it returns NULL
if there is no vm_page_t at the specified index. Otherwise, it
returns the vm_page_t. The motivation for this change is that it
simplifies the use of radix tries in the amd64, arm64, and i386 pmap
implementations. Instead of performing a lookup before every remove,
the pmap can simply perform the remove.
Reviewed by: kib, markj
Differential Revision: https://reviews.freebsd.org/D8708
called to allocate a new page of radix trie nodes, there could be a call to
vm_radix_remove() on the same trie (of PG_CACHED pages) as the in-progress
vm_radix_insert(). With the removal of PG_CACHED pages, we can simplify
vm_radix_insert() and vm_radix_remove() by removing the flags on the root of
the trie that were used to detect this case and the code for restarting
vm_radix_insert() when it happened.
Reviewed by: kib, markj
Tested by: pho
Sponsored by: Dell EMC Isilon
Differential Revision: https://reviews.freebsd.org/D8664
These changes prevent sysctl(8) from returning proper output,
such as:
1) no output from sysctl(8)
2) erroneously returning ENOMEM with tools like truss(1)
or uname(1)
truss: can not get etype: Cannot allocate memory
there is an environment variable which shall initialize the SYSCTL
during early boot. This works for all SYSCTL types both statically and
dynamically created ones, except for the SYSCTL NODE type and SYSCTLs
which belong to VNETs. A new flag, CTLFLAG_NOFETCH, has been added to
be used in the case a tunable sysctl has a custom initialisation
function allowing the sysctl to still be marked as a tunable. The
kernel SYSCTL API is mostly the same, with a few exceptions for some
special operations like iterating childrens of a static/extern SYSCTL
node. This operation should probably be made into a factored out
common macro, hence some device drivers use this. The reason for
changing the SYSCTL API was the need for a SYSCTL parent OID pointer
and not only the SYSCTL parent OID list pointer in order to quickly
generate the sysctl path. The motivation behind this patch is to avoid
parameter loading cludges inside the OFED driver subsystem. Instead of
adding special code to the OFED driver subsystem to post-load tunables
into dynamically created sysctls, we generalize this in the kernel.
Other changes:
- Corrected a possibly incorrect sysctl name from "hw.cbb.intr_mask"
to "hw.pcic.intr_mask".
- Removed redundant TUNABLE statements throughout the kernel.
- Some minor code rewrites in connection to removing not needed
TUNABLE statements.
- Added a missing SYSCTL_DECL().
- Wrapped two very long lines.
- Avoid malloc()/free() inside sysctl string handling, in case it is
called to initialize a sysctl from a tunable, hence malloc()/free() is
not ready when sysctls from the sysctl dataset are registered.
- Bumped FreeBSD version to indicate SYSCTL API change.
MFC after: 2 weeks
Sponsored by: Mellanox Technologies
To reduce the diff struct pcu.cnt field was not renamed, so
PCPU_OP(cnt.field) is still used. pc_cnt and pcpu are also used in
kvm(3) and vmstat(8). The goal was to not affect externally used KPI.
Bump __FreeBSD_version_ in case some out-of-tree module/code relies on the
the global cnt variable.
Exp-run revealed no ports using it directly.
No objection from: arch@
Sponsored by: EMC / Isilon Storage Division
reclaim the last preexisting cached page in the object, resulting in a call
to vdrop(). Detect this scenario so that the vnode's hold count is
correctly maintained. Otherwise, we panic.
Reported by: scottl
Tested by: pho
Discussed with: attilio, jeff, kib
for nodes used in vm_radix.
On architectures supporting direct mapping, also avoid to pre-allocate
the KVA for such nodes.
In order to do so make the operations derived from vm_radix_insert()
to fail and handle all the deriving failure of those.
vm_radix-wise introduce a new function called vm_radix_replace(),
which can replace a leaf node, already present, with a new one,
and take into account the possibility, during vm_radix_insert()
allocation, that the operations on the radix trie can recurse.
This means that if operations in vm_radix_insert() recursed
vm_radix_insert() will start from scratch again.
Sponsored by: EMC / Isilon storage division
Reviewed by: alc (older version)
Reviewed by: jeff
Tested by: pho, scottl
functions, reverse the numbering scheme for the levels. The highest
numbered level in the tree now appears near the root instead of the leaves.
Sponsored by: EMC / Isilon Storage Division
change the way that these functions ascend the tree when the search for a
matching leaf fails at an interior node. Rather than returning to the root
of the tree and repeating the lookup with an updated key, maintain a stack
of interior nodes that were visited during the descent and use that stack
to resume the lookup at the closest ancestor that might have a matching
descendant.
Sponsored by: EMC / Isilon Storage Division
Reviewed by: attilio
Tested by: pho
the number of interior nodes, we have previously created a level zero
interior node at the root of every non-empty trie, even when that node is
not strictly necessary, i.e., it has only one child. This change is the
second (and final) step in eliminating those unnecessary level zero interior
nodes. Specifically, it updates the deletion and insertion functions so
that they do not require a level zero interior node at the root of the trie.
For a "buildworld" workload, this change results in a 16.8% reduction in the
number of interior nodes allocated and a similar reduction in the average
execution time for lookup functions. For example, the average execution
time for a call to vm_radix_lookup_ge() is reduced by 22.9%.
Reviewed by: attilio, jeff (an earlier version)
Sponsored by: EMC / Isilon Storage Division
