regex: -G<pattern> feeds a non NUL-terminated string to regexec() and fails
When our pickaxe code feeds file contents to regexec(), it implicitly
assumes that the file contents are read into implicitly NUL-terminated
buffers (i.e. that we overallocate by 1, appending a single '\0').
This is not so.
In particular when the file contents are simply mmap()ed, we can be
virtually certain that the buffer is preceding uninitialized bytes, or
invalid pages.
Note that the test we add here is known to be flakey: we simply cannot
know whether the byte following the mmap()ed ones is a NUL or not.
Typically, on Linux the test passes. On Windows, it fails virtually
every time due to an access violation (that's a segmentation fault for
you Unix-y people out there). And Windows would be correct: the
regexec() call wants to operate on a regular, NUL-terminated string,
there is no NUL in the mmap()ed memory range, and it is undefined
whether the next byte is even legal to access.
When run with --valgrind it demonstrates quite clearly the breakage, of
course.
Being marked with `test_expect_failure`, this test will sometimes be
declare "TODO fixed", even if it only passes by mistake.
This test case represents a Minimal, Complete and Verifiable Example of
a breakage reported by Chris Sidi.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-21 21:23:22 +03:00
|
|
|
#!/bin/sh
|
|
|
|
|
#
|
|
|
|
|
# Copyright (c) 2016 Johannes Schindelin
|
|
|
|
|
#
|
|
|
|
|
|
|
|
|
|
test_description='Pickaxe options'
|
|
|
|
|
|
|
|
|
|
. ./test-lib.sh
|
|
|
|
|
|
|
|
|
|
test_expect_success setup '
|
|
|
|
|
test_commit initial &&
|
|
|
|
|
printf "%04096d" 0 >4096-zeroes.txt &&
|
|
|
|
|
git add 4096-zeroes.txt &&
|
|
|
|
|
test_tick &&
|
|
|
|
|
git commit -m "A 4k file"
|
|
|
|
|
'
|
2017-08-09 17:15:10 +03:00
|
|
|
|
|
|
|
|
# OpenBSD only supports up to 255 repetitions, so repeat twice for 64*64=4096.
|
2016-09-21 21:24:14 +03:00
|
|
|
test_expect_success '-G matches' '
|
2017-08-09 17:15:10 +03:00
|
|
|
git diff --name-only -G "^(0{64}){64}$" HEAD^ >out &&
|
regex: -G<pattern> feeds a non NUL-terminated string to regexec() and fails
When our pickaxe code feeds file contents to regexec(), it implicitly
assumes that the file contents are read into implicitly NUL-terminated
buffers (i.e. that we overallocate by 1, appending a single '\0').
This is not so.
In particular when the file contents are simply mmap()ed, we can be
virtually certain that the buffer is preceding uninitialized bytes, or
invalid pages.
Note that the test we add here is known to be flakey: we simply cannot
know whether the byte following the mmap()ed ones is a NUL or not.
Typically, on Linux the test passes. On Windows, it fails virtually
every time due to an access violation (that's a segmentation fault for
you Unix-y people out there). And Windows would be correct: the
regexec() call wants to operate on a regular, NUL-terminated string,
there is no NUL in the mmap()ed memory range, and it is undefined
whether the next byte is even legal to access.
When run with --valgrind it demonstrates quite clearly the breakage, of
course.
Being marked with `test_expect_failure`, this test will sometimes be
declare "TODO fixed", even if it only passes by mistake.
This test case represents a Minimal, Complete and Verifiable Example of
a breakage reported by Chris Sidi.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-21 21:23:22 +03:00
|
|
|
test 4096-zeroes.txt = "$(cat out)"
|
|
|
|
|
'
|
|
|
|
|
|
pickaxe: fix segfault with '-S<...> --pickaxe-regex'
'git {log,diff,...} -S<...> --pickaxe-regex' can segfault as a result
of out-of-bounds memory reads.
diffcore-pickaxe.c:contains() looks for all matches of the given regex
in a buffer in a loop, advancing the buffer pointer to the end of the
last match in each iteration. When we switched to REG_STARTEND in
b7d36ffca (regex: use regexec_buf(), 2016-09-21), we started passing
the size of that buffer to the regexp engine, too. Unfortunately,
this buffer size is never updated on subsequent iterations, and as the
buffer pointer advances on each iteration, this "bufptr+bufsize"
points past the end of the buffer. This results in segmentation
fault, if that memory can't be accessed. In case of 'git log' it can
also result in erroneously listed commits, if the memory past the end
of buffer is accessible and happens to contain data matching the
regex.
Reduce the buffer size on each iteration as the buffer pointer is
advanced, thus maintaining the correct end of buffer location.
Furthermore, make sure that the buffer pointer is not dereferenced in
the control flow statements when we already reached the end of the
buffer.
The new test is flaky, I've never seen it fail on my Linux box even
without the fix, but this is expected according to db5dfa3 (regex:
-G<pattern> feeds a non NUL-terminated string to regexec() and fails,
2016-09-21). However, it did fail on Travis CI with the first (and
incomplete) version of the fix, and based on that commit message I
would expect the new test without the fix to fail most of the time on
Windows.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-03-18 21:24:08 +03:00
|
|
|
test_expect_success '-S --pickaxe-regex' '
|
|
|
|
|
git diff --name-only -S0 --pickaxe-regex HEAD^ >out &&
|
|
|
|
|
verbose test 4096-zeroes.txt = "$(cat out)"
|
|
|
|
|
'
|
|
|
|
|
|
regex: -G<pattern> feeds a non NUL-terminated string to regexec() and fails
When our pickaxe code feeds file contents to regexec(), it implicitly
assumes that the file contents are read into implicitly NUL-terminated
buffers (i.e. that we overallocate by 1, appending a single '\0').
This is not so.
In particular when the file contents are simply mmap()ed, we can be
virtually certain that the buffer is preceding uninitialized bytes, or
invalid pages.
Note that the test we add here is known to be flakey: we simply cannot
know whether the byte following the mmap()ed ones is a NUL or not.
Typically, on Linux the test passes. On Windows, it fails virtually
every time due to an access violation (that's a segmentation fault for
you Unix-y people out there). And Windows would be correct: the
regexec() call wants to operate on a regular, NUL-terminated string,
there is no NUL in the mmap()ed memory range, and it is undefined
whether the next byte is even legal to access.
When run with --valgrind it demonstrates quite clearly the breakage, of
course.
Being marked with `test_expect_failure`, this test will sometimes be
declare "TODO fixed", even if it only passes by mistake.
This test case represents a Minimal, Complete and Verifiable Example of
a breakage reported by Chris Sidi.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-21 21:23:22 +03:00
|
|
|
test_done
|
