When write_or_die() sees EPIPE, it treats it specially by
converting it into a SIGPIPE death. We obviously cannot
ignore it, as the write has failed and the caller expects us
to die. But likewise, we cannot just call die(), because
printing any message at all would be a nuisance during
normal operations.
However, this is a problem if write_or_die() is called from
a thread. Our raised signal ends up killing the whole
process, when logically we just need to kill the thread
(after all, if we are ignoring SIGPIPE, there is good reason
to think that the main thread is expecting to handle it).
Inside an async thread, the die() code already does the
right thing, because we use our custom die_async() routine,
which calls pthread_join(). So ideally we would piggy-back
on that, and simply call:
die_quietly_with_code(141);
or similar. But refactoring the die code to do this is
surprisingly non-trivial. The die_routines themselves handle
both printing and the decision of the exit code. Every one
of them would have to be modified to take new parameters for
the code, and to tell us to be quiet.
Instead, we can just teach write_or_die() to check for the
async case and handle it specially. We do have to build an
interface to abstract the async exit, but it's simple and
self-contained. If we had many call-sites that wanted to do
this die_quietly_with_code(), this approach wouldn't scale
as well, but we don't. This is the only place where do this
weird exit trick.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We write each line of a new packed-refs file individually
using a write() syscall (and sometimes 2, if the ref is
peeled). Since each line is only about 50-100 bytes long,
this creates a lot of system call overhead.
We can instead open a stdio handle around our descriptor and
use fprintf to write to it. The extra buffering is not a
problem for us, because nobody will read our new packed-refs
file until we call commit_lock_file (by which point we have
flushed everything).
On a pathological repository with 8.5 million refs, this
dropped the time to run `git pack-refs` from 20s to 6s.
Signed-off-by: Jeff King <peff@peff.net>
Reviewed-by: Michael Haggerty <mhagger@alum.mit.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The write_or_die function will always die on an error,
including EPIPE. However, it currently treats EPIPE
specially by suppressing any error message, and by exiting
with exit code 0.
Suppressing the error message makes some sense; a pipe death
may just be a sign that the other side is not interested in
what we have to say. However, exiting with a successful
error code is not a good idea, as write_or_die is frequently
used in cases where we want to be careful about having
written all of the output, and we may need to signal to our
caller that we have done so (e.g., you would not want a push
whose other end has hung up to report success).
This distinction doesn't typically matter in git, because we
do not ignore SIGPIPE in the first place. Which means that
we will not get EPIPE, but instead will just die when we get
a SIGPIPE. But it's possible for a default handler to be set
by a parent process, or for us to add a callsite inside one
of our few SIGPIPE-ignoring blocks of code.
This patch converts write_or_die to actually raise SIGPIPE
when we see EPIPE, rather than exiting with zero. This
brings the behavior in line with the "normal" case that we
die from SIGPIPE (and any callers who want to check why we
died will see the same thing). We also give the same
treatment to other related functions, including
write_or_whine_pipe and maybe_flush_or_die.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
check to compat
Commit b2f5e268 (Windows: Work around an oddity when a pipe with no reader
is written to) introduced a check for EINVAL after fflush() to fight
spurious "Invalid argument" errors on Windows when a pipe was broken. But
this check may hide real errors on systems that do not have the this odd
behavior. Introduce an fflush wrapper in compat/mingw.* so that the treatment
is only applied on Windows.
Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change calls to die(..., strerror(errno)) to use the new die_errno().
In the process, also make slight style adjustments: at least state
_something_ about the function that failed (instead of just printing
the pathname), and put paths in single quotes.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A few compat/* layer functions call these functions, but we would really
want to keep them thin, without depending too much on the libgit proper.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
On Windows, write() is implemented using WriteFile(). After the reader
closed its end of the pipe, the first WriteFile() returns
ERROR_BROKEN_PIPE (which translates to EPIPE), subsequent WriteFile()s
return ERROR_NO_DATA, which is translated to EINVAL.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
This means that we can depend on packs always being stable on disk,
simplifying a lot of the object serialization worries. And unlike loose
objects, serializing pack creation IO isn't going to be a performance
killer.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
xread() and xwrite() return ssize_t values as their native POSIX
counterparts read(2) and write(2).
To be consistent, read_in_full() and write_in_full() should also return
ssize_t values.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch arose from a discussion started by Jim Meyering's patch
whose intention was to provide better diagnostics for failed writes.
Linus proposed a better way to do things, which also had the added
benefit that adding a fflush() to git-log-* operations and incremental
git-blame operations could improve interactive respose time feel, at
the cost of making things a bit slower when we aren't piping the
output to a downstream program.
This patch skips the fflush() calls when stdout is a regular file, or
if the environment variable GIT_FLUSH is set to "0". This latter can
speed up a command such as:
GIT_FLUSH=0 strace -c -f -e write time git-rev-list HEAD | wc -l
a tiny amount.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It received the return value from xwrite() in a size_t variable
'written' and expected comparison with 0 would catch an error
from xwrite().
Signed-off-by: Junio C Hamano <junkio@cox.net>
Originally I introduced read_or_die for the purpose of reading
the pack header and trailer, and I was too lazy to print proper
error messages.
Linus Torvalds <torvalds@osdl.org>:
> For a read error, at the very least you have to say WHICH FILE
> couldn't be read, because it's usually a matter of some file just
> being too short, not some system-wide problem.
and of course Linus is right. Make it so.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
The "read_or_die()" function would silently NOT die for a partial read,
and since it was of type "void" it obviously couldn't even return the
partial number of bytes read.
IOW, it was totally broken. This hopefully fixes it up.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
With the new-and-improved write_in_full() semantics, where a partial write
simply always returns a real error (and always sets 'errno' when that
happens, including for the disk full case), a lot of the callers of
write_in_full() were just unnecessarily complex.
In particular, there's no reason to ever check for a zero length or
return: if the length was zero, we'll return zero, otherwise, if a disk
full resulted in the actual write() system call returning zero the
write_in_full() logic would have correctly turned that into a negative
return value, with 'errno' set to ENOSPC.
I really wish every "write_in_full()" user would just check against "<0"
now, but this fixes the nasty and stupid ones.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Unfortunately, while {read,write}_in_full do take into account
zero-sized reads/writes; their die and whine variants do not.
I have a repository where there are zero-sized files in
the history that was triggering these things.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
We have a number of badly checked write() calls. Often we are
expecting write() to write exactly the size we requested or fail,
this fails to handle interrupts or short writes. Switch to using
the new write_in_full(). Otherwise we at a minimum need to check
for EINTR and EAGAIN, where this is appropriate use xwrite().
Note, the changes to config handling are much larger and handled
in the next patch in the sequence.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
We have a number of badly checked read() calls. Often we are
expecting read() to read exactly the size we requested or fail, this
fails to handle interrupts or short reads. Add a read_in_full()
providing those semantics. Otherwise we at a minimum need to check
for EINTR and EAGAIN, where this is appropriate use xread().
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
We recently introduced a write_in_full() which would either write
the specified object or emit an error message and fail. In order
to fix the read side we now want to introduce a read_in_full()
but without an error emit. This patch cleans up the naming
of this family of calls:
1) convert the existing write_or_whine() to write_or_whine_pipe()
to better indicate its pipe specific nature,
2) convert the existing write_in_full() calls to write_or_whine()
to better indicate its nature,
3) introduce a write_in_full() providing a write or fail semantic,
and
4) convert write_or_whine() and write_or_whine_pipe() to use
write_in_full().
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
* sp/mmap: (27 commits)
Spell default packedgitlimit slightly differently
Increase packedGit{Limit,WindowSize} on 64 bit systems.
Update packedGit config option documentation.
mmap: set FD_CLOEXEC for file descriptors we keep open for mmap()
pack-objects: fix use of use_pack().
Fix random segfaults in pack-objects.
Cleanup read_cache_from error handling.
Replace mmap with xmmap, better handling MAP_FAILED.
Release pack windows before reporting out of memory.
Default core.packdGitWindowSize to 1 MiB if NO_MMAP.
Test suite for sliding window mmap implementation.
Create pack_report() as a debugging aid.
Support unmapping windows on 'temporary' packfiles.
Improve error message when packfile mmap fails.
Ensure core.packedGitWindowSize cannot be less than 2 pages.
Load core configuration in git-verify-pack.
Fully activate the sliding window pack access.
Unmap individual windows rather than entire files.
Document why header parsing won't exceed a window.
Loop over pack_windows when inflating/accessing data.
...
Conflicts:
cache.h
pack-check.c
When passing the revisions list to pack-objects we do not check for
errors nor short writes. Introduce a new write_in_full which will
handle short writes and report errors to the caller. Use this to
short cut the send on failure, allowing us to wait for and report
the child in case the failure is its fault.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Like write_or_die read_or_die reads the entire length requested
or it kills the current process with a die call.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Now if GIT_TRACE is set to an integer value greater than 1
and lower than 10, we interpret this as an open fd value
and we trace into it. Note that this behavior is not
compatible with the previous one.
We also trace whole messages using one write(2) call to
make sure messages from processes do net get mixed up in
the middle.
It's now possible to run the tests like this:
GIT_TRACE=9 make test 9>/var/tmp/trace.log
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
The little helper write_or_die() won't come back with bad news about
full disks or broken pipes. It either succeeds or terminates the
program, making additional error handling unnecessary.
This patch adds the new function and uses it to replace two similar
ones (the one in tar-tree originally has been copied from cat-file
btw.). I chose to add the fd parameter which both lacked to make
write_or_die() just as flexible as write() and thus suitable for
lib-ification.
There is a regression: error messages emitted by this function don't
show the program name, while the replaced two functions did. That's
acceptable, I think; a lot of other functions do the same.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
