fixed (many due to changing semantics in other parts of the kernel and not
the original author's fault), including one critical one: unionfs could
cause UFS corruption in the fronting store due to calling VOP_OPEN for
writing without turning on vmio for the UFS vnode.
Most of the bugs were related to semantics changes in VOP calls, lock
ordering problems (causing deadlocks), improper handling of a read-only
backing store (such as an NFS mount), improper referencing and locking
of vnodes, not using real struct locks for vnode locking, not using
recursive locks when accessing the fronting store, and things like that.
New functionality has been added: unionfs now has mmap() support, but
only partially tested, and rename has been enhanced considerably.
There are still some things that unionfs cannot do. You cannot
rename a directory without confusing unionfs, and there are issues
with softlinks, hardlinks, and special files. unionfs mostly doesn't
understand them (and never did).
There are probably still panic situations, but hopefully no where near
as many as before this commit.
The unionfs in this commit has been tested overlayed on /usr/src
(backing /usr/src being a read-only NFS mount, fronting /usr/src being
a local filesystem). kernel builds have been tested, buildworld is
undergoing testing. More testing is necessary.
lock specifications in kern/vnode_if.src. At present, this do not
distinguish between exclusive and shared locks, and the kernel is so full
of bugs in this area that running with auto-generation of assertions
enabled makes DEBUG_VFS_LOCKS totally useless for anybody that has used it
for anything prior to outputting automated assertions. Due to this, I made
vnode_if.sh only output locking assertions if you have the environment
variable DEBUG_ALL_VFS_LOCKS set to "YES". In order to actually use the
assertions, you need to also add "options DEBUG_VFS_LOCKS" to your kernel
config file.
Urged to commit by: phk
- Split out the prototypes, externs and struct decls from if_epreg.h into
if_epvar.h.
- Add support for MCA based Etherlink III (3c529) devices.
None of this code is used right now; the old if_ep driver is still
in place and used.
I will eventually get around to converting if_ep_isa.c to newbus once I've
had a talk with Peter and DFR about the DEVICE_IDENTIFY() method.
I have tested this code on my PS/2. It works. I would like EISA and ISA
testers since my example hardware hasn't arrived yet.
Add:
dev/ep/if_ep.c optional ep
dev/ep/if_ep_isa.c optional ep isa
dev/ep/if_ep_eisa.c optional ep eisa
dev/ep/if_ep_mca.c optional ep mca
dev/ep/if_ep_pccard.c optional ep card
to sys/conf/files
Remove:
i386/eisa/3c5x9.c optional ep
i386/isa/if_ep.c optional ep
from sys/i386/conf/files.i386
PCCARD testers wanted!
I will switch off and cvs rm the old driver in favor of this copy once
I've had positive feedback or have the hardware to verify that it works.
have been there in the first place. A GENERIC kernel shrinks almost 1k.
Add a slightly different safetybelt under nostop for tty drivers.
Add some missing FreeBSD tags
fields in struct cdevsw:
d_stop moved to struct tty.
d_reset already unused.
d_devtotty linkage now provided by dev_t->si_tty.
These fields will be removed from struct cdevsw together with
d_params and d_maxio Real Soon Now.
The changes in this patch consist of:
initialize dev->si_tty in *_open()
initialize tty->t_stop
remove devtotty functions
rename ttpoll to ttypoll
a few adjustments to these changes in the generic code
a bump of __FreeBSD_version
add a couple of FreeBSD tags
This means that we will not have to have a bpf and a non-bpf version
of our driver modules.
This does not open any security hole, because the bpf core isn't loadable
The drivers left unchanged are the "cross platform" drivers where the respective
maintainers are urged to DTRT, whatever that may be.
Add a couple of missing FreeBSD tags.
for you to be told there was an error [during verbose boot].
I poked him for the fix, he poked me to get it committed.
Submitted by: Jason Young <doogie@anet-stl.com>
Enhance MIRROR code. Add a few more sanity checks and implement
a zone-based disk selector to make use of both disks when reading.
Also implement a read fail-over. If a read error occurs on one
disk, the I/O is retried on the other.
NOTE: CCD's mirroring support cannot deal with write errors properly
in regards to recovery, meaning that 'old' data under a write error may
be read non-deterministically if you reboot after a write error, and CCD
certainly cannot deal with a disk changeout. And it still can't. Use
vinum if you are really serious about mirroring. CCD basically just
implements a poor-man's mirror.
sum the total amount of I/O issued to determine when all the I/O
has completed. This fails when the EOF boundry occurs in the middle
of an I/O. Using cbp->cb_buf.b_bufsize works better.
there are stubs compiled into the kernel if BPF support is not enabled,
there aren't any problems with unresolved symbols. The modules in /modules
are compiled with BPF support enabled anyway, so the most this will do is
bloat GENERIC a little.
