| Summary: | reproducer for races of expect incomplete reads | ||
|---|---|---|---|
| Product: | gdb | Reporter: | Jan Kratochvil <jan> |
| Component: | testsuite | Assignee: | Not yet assigned to anyone <unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | enhancement | CC: | pedro, polacek, ssbssa, vries |
| Priority: | P2 | ||
| Version: | HEAD | ||
| Target Milestone: | 7.9 | ||
| Host: | Target: | i686-fedora15-linux-gnu | |
| Build: | Last reconfirmed: | ||
| Bug Depends on: | |||
| Bug Blocks: | 12647 | ||
| Attachments: |
PASSes->FAILs after this reproducer.
gdb_test_timeout *= 20 writew: A different reproducer - delaying the GDB writes. |
||
*** Bug 12647 has been marked as a duplicate of this bug. *** Created attachment 5650 [details]
gdb_test_timeout *= 20
And be careful with multiarch, /usr/local/lib/read1.so has only single arch.
Brilliant! CVSROOT: /cvs/src Module name: src Changes by: palves@sourceware.org 2011-05-02 17:52:28 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.base: completion.exp Log message: 2011-05-02 Pedro Alves <pedro@codesourcery.com> PR testsuite/12649 Fix races. * gdb.base/completion.exp: Remove all sleep calls. Remove unnecessary regexs. Don't explicitly expect anything after the prompt. Eat the prompt if necessary. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.2687&r2=1.2688 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.base/completion.exp.diff?cvsroot=src&r1=1.50&r2=1.51 Just a little status report, where are we at. These tests were already race-fixed: * gdb.base/commands.exp * gdb.base/completion.exp * gdb.base/pointers.exp * gdb.base/pr10179.exp * gdb.base/setshow.exp * gdb.cp/cpcompletion.exp * gdb.cp/static-print-quit.exp * gdb.mi/mi-basics.exp * gdb.mi/mi-nonstop-exit.exp * gdb.mi/mi-nonstop.exp * gdb.mi/mi-ns-stale-regcache.exp * gdb.mi/mi-nsthrexec.exp Remaining: * gdb.mi/mi-nsintrall.exp * gdb.mi/mi-nsmoribund.exp * gdb.mi/mi-return.exp * gdb.mi/mi-var-display.exp * gdb.mi/mi2-basics.exp * gdb.mi/mi2-console.exp * gdb.mi/mi2-return.exp * gdb.mi/mi2-var-display.exp * gdb.python/py-evthreads.exp * gdb.python/python.exp Created attachment 6039 [details] writew: A different reproducer - delaying the GDB writes. It could reproduce: FAIL: gdb.mi/mi-nonstop.exp: breakpoint at marker from: [patch] testsuite: MI: racy results in async mode #2 http://sourceware.org/ml/gdb-patches/2011-10/msg00797.html CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2011-11-25 23:50:16 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.reverse: i386-sse-reverse.exp Log message: gdb/testsuite/ PR testsuite/12649 * gdb.reverse/i386-sse-reverse.exp (continue to end of sse_test #2): Wrap send_gdb into a new gdb_test. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.2946&r2=1.2947 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.reverse/i386-sse-reverse.exp.diff?cvsroot=src&r1=1.6&r2=1.7 CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2011-12-02 12:43:30 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.trace: collection.exp tspeed.exp tstatus.exp Log message: gdb/testsuite/ PR testsuite/12649 * gdb.trace/collection.exp (gdb_collect_return_test): Import gdb_prompt. Protect gdb_test_multiple by final $gdb_prompt match. * gdb.trace/tspeed.exp (gdb_fast_trace_speed_test): Likewise. * gdb.trace/tstatus.exp (run_trace_experiment): Likewise. (test_tracepoints): Likewise. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.2955&r2=1.2956 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.trace/collection.exp.diff?cvsroot=src&r1=1.28&r2=1.29 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.trace/tspeed.exp.diff?cvsroot=src&r1=1.1&r2=1.2 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.trace/tstatus.exp.diff?cvsroot=src&r1=1.1&r2=1.2 CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2011-12-10 23:05:08 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.trace: ftrace.exp Log message: gdb/testsuite/ PR testsuite/12649 * gdb.trace/ftrace.exp (test_fast_tracepoints): Import gdb_prompt. Protect gdb_test_multiple by final $gdb_prompt match. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.2975&r2=1.2976 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.trace/ftrace.exp.diff?cvsroot=src&r1=1.1&r2=1.2 CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2012-05-15 13:08:58 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.base: dprintf.exp Log message: gdb/testsuite/ PR testsuite/12649 * gdb.base/dprintf.exp: Replace gdb_run_cmd by runto main. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.3189&r2=1.3190 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.base/dprintf.exp.diff?cvsroot=src&r1=1.1&r2=1.2 CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2012-07-04 14:57:24 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.base: dprintf.exp Log message: gdb/testsuite/ PR 12649 * gdb.base/dprintf.exp (Set dprintf style to agent): Add missing gdb_test_multiple $gdb_prompt terminator. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.3274&r2=1.3275 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.base/dprintf.exp.diff?cvsroot=src&r1=1.4&r2=1.5 CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2013-05-21 15:00:33 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.mi: mi-dprintf.exp Log message: gdb/testsuite/ PR testsuite/12649 * gdb.mi/mi-dprintf.exp (mi_continue_dprintf): Fix expect strings for racy matches. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.3669&r2=1.3670 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.mi/mi-dprintf.exp.diff?cvsroot=src&r1=1.1&r2=1.2 CVSROOT: /cvs/src Module name: src Changes by: jkratoch@sourceware.org 2013-05-24 15:37:25 Modified files: gdb/testsuite : ChangeLog gdb/testsuite/gdb.mi: mi-dprintf.exp Log message: gdb/testsuite/ PR testsuite/12649 * gdb.mi/mi-dprintf.exp (mi_continue_dprintf) (mi 2nd dprintf): Replace $mi_gdb_prompt expectation by mi_expect_stop. (mi 1st dprintf, agent, mi 2nd dprintf, agent) (mi info dprintf second time): Replace them by mi_send_resuming_command and mi_expect_stop. Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/ChangeLog.diff?cvsroot=src&r1=1.3680&r2=1.3681 http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.mi/mi-dprintf.exp.diff?cvsroot=src&r1=1.2&r2=1.3 Unassigning. This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "gdb and binutils".
The branch, master has been updated
via 2a31c6236de308f7109720230e70b96c5d5b966a (commit)
from 469a34932ae4208240e98c843b01ea1f00892137 (commit)
Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.
- Log -----------------------------------------------------------------
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=2a31c6236de308f7109720230e70b96c5d5b966a
commit 2a31c6236de308f7109720230e70b96c5d5b966a
Author: Pedro Alves <palves@redhat.com>
Date: Wed Aug 20 18:55:54 2014 +0100
Integrate PR 12649's race detector directly in the testsuite machinery
This integrates Jan Kratochvil's nice race reproducer from PR
testsuite/12649 into the testsuite infrustructure directly.
With this, one only has to do either 'make check-read1' or 'make check
READ1="1"' to preload the read1.so library into expect.
Currently only enabled for glibc/GNU systems, and if
build==host==target.
gdb/testsuite/ChangeLog:
* Makefile.in (EXTRA_RULES, CC): New variables, get from
configure.
(EXPECT): Handle READ1 being set.
(all): Depend on EXTRA_RULES.
(check-read1, expect-read1, read1.so, read1): New rules.
* README (Testsuite Parameters): Document the READ1 make variable.
(Race detection): New section.
* configure: Regenerate.
* configure.ac: If build==host==target, and running under a
GNU/glibc system, add read1 to the extra Makefile rules.
(EXTRA_RULES): AC_SUBST it.
* lib/read1.c: New file.
gdb/ChangeLog:
* Makefile.in (check-read1): New rule.
-----------------------------------------------------------------------
Summary of changes:
gdb/ChangeLog | 4 ++
gdb/Makefile.in | 8 ++++
gdb/testsuite/ChangeLog | 16 +++++++
gdb/testsuite/Makefile.in | 44 ++++++++++++++++++--
gdb/testsuite/README | 41 ++++++++++++++++++
gdb/testsuite/configure | 10 ++++
gdb/testsuite/configure.ac | 9 ++++
.../expect-read1.c => testsuite/lib/read1.c} | 16 ++++----
8 files changed, 136 insertions(+), 12 deletions(-)
copy gdb/{contrib/expect-read1.c => testsuite/lib/read1.c} (80%)
(In reply to Sourceware Commits from comment #15) > This is an automated email from the git hooks/post-receive script. It was > generated because a ref change was pushed to the repository containing > the project "gdb and binutils". > > The branch, master has been updated > via 2a31c6236de308f7109720230e70b96c5d5b966a (commit) > from 469a34932ae4208240e98c843b01ea1f00892137 (commit) > > Those revisions listed above that are new to this repository have > not appeared on any other notification email; so we list those > revisions in full, below. > > - Log ----------------------------------------------------------------- > https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git; > h=2a31c6236de308f7109720230e70b96c5d5b966a > > commit 2a31c6236de308f7109720230e70b96c5d5b966a > Author: Pedro Alves <palves@redhat.com> > Date: Wed Aug 20 18:55:54 2014 +0100 > > Integrate PR 12649's race detector directly in the testsuite machinery > > This integrates Jan Kratochvil's nice race reproducer from PR > testsuite/12649 into the testsuite infrustructure directly. > > With this, one only has to do either 'make check-read1' or 'make check > READ1="1"' to preload the read1.so library into expect. > > Currently only enabled for glibc/GNU systems, and if > build==host==target. > > gdb/testsuite/ChangeLog: > > * Makefile.in (EXTRA_RULES, CC): New variables, get from > configure. > (EXPECT): Handle READ1 being set. > (all): Depend on EXTRA_RULES. > (check-read1, expect-read1, read1.so, read1): New rules. > * README (Testsuite Parameters): Document the READ1 make variable. > (Race detection): New section. > * configure: Regenerate. > * configure.ac: If build==host==target, and running under a > GNU/glibc system, add read1 to the extra Makefile rules. > (EXTRA_RULES): AC_SUBST it. > * lib/read1.c: New file. > > gdb/ChangeLog: > > * Makefile.in (check-read1): New rule. > > ----------------------------------------------------------------------- > > Summary of changes: > gdb/ChangeLog | 4 ++ > gdb/Makefile.in | 8 ++++ > gdb/testsuite/ChangeLog | 16 +++++++ > gdb/testsuite/Makefile.in | 44 > ++++++++++++++++++-- > gdb/testsuite/README | 41 ++++++++++++++++++ > gdb/testsuite/configure | 10 ++++ > gdb/testsuite/configure.ac | 9 ++++ > .../expect-read1.c => testsuite/lib/read1.c} | 16 ++++---- > 8 files changed, 136 insertions(+), 12 deletions(-) > copy gdb/{contrib/expect-read1.c => testsuite/lib/read1.c} (80%) Can this be closed? (In reply to Hannes Domani from comment #16) > Can this be closed? It can, thanks for the reminder. |
Created attachment 5649 [details] PASSes->FAILs after this reproducer. Excercise any incomplete reads handling in the testsuite by simulating read() always returns just 1 character. #define _GNU_SOURCE 1 #include <dlfcn.h> #include <unistd.h> #include <fcntl.h> #include <stdlib.h> ssize_t read (int fd, void *buf, size_t count) { static ssize_t (*read2) (int fd, void *buf, size_t count) = NULL; if (read2 == NULL) { unsetenv ("LD_PRELOAD"); read2 = dlsym (RTLD_NEXT, "read"); } if (count > 0 && isatty (fd) >= 1) count = 1; return read2 (fd, buf, count); } gcc -o /usr/local/lib/read1.so read1.c -Wall -g -shared -fPIC /usr/local/bin/runtest: #! /bin/sh export EXPECT=/usr/local/bin/expect-read1 exec /usr/bin/runtest "$@" /usr/local/bin/expect-read1: #! /bin/sh export LD_PRELOAD=/usr/local/lib/read1.so exec /usr/bin/expect "$@" -# of expected passes 19875 +# of expected passes 19701 -# of unexpected failures 69 +# of unexpected failures 173 +# of unknown successes 9 +# of known failures 71 +# of unresolved testcases 9 (Some of the differences in the attachment may be unrelated to this Bug.)