Re: [PATCH] timeout: fix failure if timeout's parent has ignored SIGCHLD
I'm a bit worried about doing the following in this patch. It seems to work but I can see ignoring CHLD being problematic for some shells/situations. I'll probably push without the test. --- a/tests/misc/timeout +++ b/tests/misc/timeout @@ -23,6 +23,7 @@ fi . $srcdir/test-lib.sh +trap '' CHLD # ensure `timeout` is immune to parent's SIGCHLD handler
[PATCH] sort: fix failure if sort's parent has ignored SIGCHLD
The same issue as with the previous `timeout` patch. The attached includes at test, but I'll probably remove it because of the previously mentioned concerns. cheers, Pádraig.
Re: stty & stty-row-col test failures
I'm going with this unless there are objections >From bb0ae98acca9fe2b27eb5ccadf57d4842cab5d8f Mon Sep 17 00:00:00 2001 From: =?utf-8?q?P=C3=A1draig=20Brady?= Date: Tue, 8 Dec 2009 08:00:37 + Subject: [PATCH] tests: fix stty-row-col failure on small fixed terminals * tests/misc/stty-row-col: Linux virtual consoles at least, issue an error if you try to increase their size, so skip the test if we can't increase the dimensions of the tty by 1 cell. Reported by Matthew Burgess. --- tests/misc/stty-row-col |9 + 1 files changed, 9 insertions(+), 0 deletions(-) diff --git a/tests/misc/stty-row-col b/tests/misc/stty-row-col index 5efc7dd..943565f 100755 --- a/tests/misc/stty-row-col +++ b/tests/misc/stty-row-col @@ -60,6 +60,15 @@ set $tests saved_size=`stty size` && test -n "$saved_size" \ || skip_test_ "can't get window size" +# Linux virtual consoles issue an error if you +# try to increase their size. So skip in that case. +if ! test "x$saved_size" = "x0 0"; then + srow=`echo $saved_size | cut -d ' ' -f1` + scol=`echo $saved_size | cut -d ' ' -f2` + stty rows `expr $srow + 1` cols `expr $scol + 1` || +skip_test_ "can't increase window size" +fi + while :; do test_name=$1 args=$2 -- 1.6.2.5
Re: stty & stty-row-col test failures
Pádraig Brady wrote: > I'm going with this unless there are objections > >>From bb0ae98acca9fe2b27eb5ccadf57d4842cab5d8f Mon Sep 17 00:00:00 2001 > From: =?utf-8?q?P=C3=A1draig=20Brady?= > Date: Tue, 8 Dec 2009 08:00:37 + > Subject: [PATCH] tests: fix stty-row-col failure on small fixed terminals > > * tests/misc/stty-row-col: Linux virtual consoles at least, > issue an error if you try to increase their size, so skip the > test if we can't increase the dimensions of the tty by 1 cell. > Reported by Matthew Burgess. > --- > tests/misc/stty-row-col |9 + > 1 files changed, 9 insertions(+), 0 deletions(-) > > diff --git a/tests/misc/stty-row-col b/tests/misc/stty-row-col > index 5efc7dd..943565f 100755 > --- a/tests/misc/stty-row-col > +++ b/tests/misc/stty-row-col > @@ -60,6 +60,15 @@ set $tests > saved_size=`stty size` && test -n "$saved_size" \ >|| skip_test_ "can't get window size" > > +# Linux virtual consoles issue an error if you > +# try to increase their size. So skip in that case. > +if ! test "x$saved_size" = "x0 0"; then > + srow=`echo $saved_size | cut -d ' ' -f1` > + scol=`echo $saved_size | cut -d ' ' -f2` > + stty rows `expr $srow + 1` cols `expr $scol + 1` || > +skip_test_ "can't increase window size" > +fi Hi Pádraig, Thanks for tracking that down! Would you please use "!=" in that test and $(...) rather than `...` ? if test "x$saved_size" != "x0 0"; then srow=$(echo $saved_size | cut -d ' ' -f1) scol=$(echo $saved_size | cut -d ' ' -f2) stty rows $(expr $srow + 1) cols $(expr $scol + 1) || skip_test_ "can't increase window size" fi
Re: stty & stty-row-col test failures
Pádraig Brady wrote: > Matthew Burgess wrote: >> Hi, >> >> The stty & stty-row-col tests fail for me, but I can't figure out why. I >> wonder whether it's because I'm in a VirtualBox host and it's somehow >> interfering with the tests? Anyway, I've attached the logs from a >> verbose run of both tests. If someone wouldn't mind taking a look for >> me please, I'd appreciate it. >> >> This is with Coreutils-8.1 on a Kubuntu 9.10 guest under a >> VirtualBox-3.1.0 VM on a Windows XP host. >> >> If you need any more information or debug output then I'll be more than >> happy to provide it. > > I asked for a kernel version before preparing a patch > for this previously reported issue: > http://lists.gnu.org/archive/html/bug-coreutils/2009-11/msg00059.html > Since you've mentioned Karmic I'm going to apply the following. > > cheers, > Pádraig. > > diff --git a/tests/misc/stty b/tests/misc/stty > index cb3fee5..e8b2a9a 100755 > --- a/tests/misc/stty > +++ b/tests/misc/stty > @@ -63,8 +63,10 @@ options=`stty -a | tr -s ';' '\n' | sed "s/^ > //;$sed_del;s/-//g"` > for opt in $options; do ># `stty parenb' and `stty -parenb' fail with this message ># stty: standard input: unable to perform all requested operations > - # on Linux 2.2.0-pre4 kernels, so skip those tests. > - test $opt = parenb && continue > + # on Linux 2.2.0-pre4 kernels. Also since Linux 2.6.31 the > + # other serial control settings give the same error. So skip them. > + case $opt in parenb|parodd|cstopb|crtscts) continue;; esac Thanks! This has been a long-standing intermittent problem.
Re: stty & stty-row-col test failures
Jim Meyering wrote: > Thanks for tracking that down! > Would you please use "!=" in that test > and $(...) rather than `...` ? Sure, I was wary because of the mention of sunos in the test.
Re: new snapshot available: coreutils-8.1.24-7a2b0
Matthew Burgess wrote: > I get a 403 Forbidden error trying to download the latest ss tarballs > from both meyering.net & people.redhat.com. Hi Matthew. Rats. Thanks for the heads up. I've just fixed the permissions on both. That was fall-out from my switching a shell to a umask of 077. My upload procedure depended on the prior 022 setting. Must fix. > What timescales did you envisage for the 8.2 release? It should be tomorrow or Thursday. Later if there are too many changes. > I'll try to run > through an LFS build with it prior to the release so I can report any > issues I see *before* one of your stable releases for a change :-) Thanks!
Re: [PATCH] sort: fix failure if sort's parent has ignored SIGCHLD
Pádraig Brady wrote: > The same issue as with the previous `timeout` patch. > The attached includes at test, but I'll probably remove > it because of the previously mentioned concerns. really attached this time. sorry >From 33904143eab537385ce1af58ea15c1876da402c7 Mon Sep 17 00:00:00 2001 From: =?utf-8?q?P=C3=A1draig=20Brady?= Date: Tue, 8 Dec 2009 08:48:34 + Subject: [PATCH] sort: fix failure if sort's parent has ignored SIGCHLD * src/sort.c (main): Reset the SIGCHLD handler to the default as otherwise wait() could return an error. * tests/misc/sort-compress: Set the CHLD handler to SIG_IGN to ensure the sort command resets it to SIG_DFL. * NEWS: Mention the fix. --- NEWS |4 src/sort.c |1 + tests/misc/sort-compress |2 ++ 3 files changed, 7 insertions(+), 0 deletions(-) diff --git a/NEWS b/NEWS index 19cca5b..528eedc 100644 --- a/NEWS +++ b/NEWS @@ -13,6 +13,10 @@ GNU coreutils NEWS-*- outline -*- the presence of the empty string argument. [bug introduced in coreutils-8.0] + sort is now immune to the signal handling of its parent. + Specifically sort now doesn't exit with an error message + if it uses helper processes for compression and its parent + ignores CHLD signals. [bug introduced in coreutils-6.9] * Noteworthy changes in release 8.1 (2009-11-18) [stable] diff --git a/src/sort.c b/src/sort.c index 8709e53..70e0bbb 100644 --- a/src/sort.c +++ b/src/sort.c @@ -3254,6 +3254,7 @@ main (int argc, char **argv) } #endif } + signal (SIGCHLD, SIG_DFL); /* Don't inherit CHLD handling from parent. */ /* The signal mask is known, so it is safe to invoke exit_cleanup. */ atexit (exit_cleanup); diff --git a/tests/misc/sort-compress b/tests/misc/sort-compress index 7e4278e..e923e64 100755 --- a/tests/misc/sort-compress +++ b/tests/misc/sort-compress @@ -23,6 +23,8 @@ fi . $srcdir/test-lib.sh +trap '' CHLD # ensure `sort` is immune to parent's SIGCHLD handler + seq -w 2000 > exp || framework_failure tac exp > in || framework_failure SORT="$abs_top_builddir/src/sort" -- 1.6.2.5
Re: stty & stty-row-col test failures
Pádraig Brady wrote: > Jim Meyering wrote: >> Thanks for tracking that down! >> Would you please use "!=" in that test >> and $(...) rather than `...` ? > > Sure, I was wary because of the mention of sunos in the test. The configure-time test for a posix shell should cover us in the tests, since each is invoked via an explicit $(SHELL) from test/check.mk: TESTS_ENVIRONMENT = \ . $(srcdir)/lang-default; \ tmp__=$$TMPDIR; test -d "$$tmp__" || tmp__=.; \ . $(srcdir)/envvar-check; \ TMPDIR=$$tmp__; export TMPDIR;\ ... $(SHELL) "$$1"; \ fi; \ ...
Re: [PATCH] timeout: fix failure if timeout's parent has ignored SIGCHLD
Pádraig Brady wrote: > I'm a bit worried about doing the following in this patch. > It seems to work but I can see ignoring CHLD being > problematic for some shells/situations. > I'll probably push without the test. Oh it looks like tests/install/trap already addresses the portability concerns I was worrying about. I'll use that method to run the test. http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=blob;f=tests/install/trap;hb=HEAD cheers, Pádraig.
Re: [PATCH] timeout: fix failure if timeout's parent has ignored SIGCHLD
Pádraig Brady wrote: > Pádraig Brady wrote: >> I'm a bit worried about doing the following in this patch. >> It seems to work but I can see ignoring CHLD being >> problematic for some shells/situations. >> I'll probably push without the test. > > Oh it looks like tests/install/trap already addresses > the portability concerns I was worrying about. > I'll use that method to run the test. > > http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=blob;f=tests/install/trap;hb=HEAD Ahhh. Good. I'd forgotten about that. Same for the test of your sort fix, I presume.
mv -v, cp -v messages should be different
Gentlemen, I object. The messages for these two commands should be different. $ cp -v f g `f' -> `g' $ mv -v f g `f' -> `g' Exactly how different etc. I leave up to you. Maybe even just => for the latter instead of ->.
Re: mv -v, cp -v messages should be different
jida...@jidanni.org wrote: > Gentlemen, I object. The messages for these two commands should be different. > $ cp -v f g > `f' -> `g' > $ mv -v f g > `f' -> `g' > Exactly how different etc. I leave up to you. Maybe even just => for the > latter instead of ->. I think it's fine from the context as is? If one was going to change it I'd have: file == file_copy file -> file_rename file => file_copy_unlink But I don't see the need TBH. cheers, Pádraig.
Re: build: distcheck: do not leave a $TMPDIR/coreutils directory behind
Jim Meyering wrote: > FYI, I should be pushing these soon, and then making a snapshot > within a couple hours: > > [PATCH 1/2] build: distcheck: do not leave a $TMPDIR/coreutils directory > behind aka, http://git.savannah.gnu.org/cgit/coreutils.git/commit/?id=ae034822c535fa5 Now that there's a public BZ mentioning the security impact, http://bugzilla.redhat.com/545439, I will note that the above change also fixes a security-related flaw. Any user running "make distcheck" with TMPDIR unset or set to a world-writable directory like /tmp, is vulnerable to arbitrary code execution. So either don't run "make distcheck", with coreutils-8.1 or earlier, or be sure that TMPDIR is not a world-writable directory. I'll add something like the above to NEWS.
Re: git coreutils 'make check' hangs
Hi Pádraig, * Pádraig Brady wrote on Tue, Dec 08, 2009 at 02:00:20AM CET: > Ralf Wildenhues wrote: > > for some time now, 'make check' in the git coreutils tree hangs for me: [...] > I think it may be a gdb issue: > http://lists.gnu.org/archive/html/bug-coreutils/2009-12/msg00025.htm > For the moment we've marked that test as "very expensive" > so that it will not be run by default. Ah, ok, I didn't see that thread. Sorry about the noise then. Thanks, Ralf
new snapshot available: coreutils-8.1.30-0f8bb
This should be the final snapshot before coreutils-8.2, due tomorrow or Thursday. coreutils snapshot: (.gz files are here, too) http://meyering.net/cu/coreutils-ss.tar.xz 4.3 MB http://meyering.net/cu/coreutils-ss.tar.xz.sig http://meyering.net/cu/coreutils-8.1.30-0f8bb.tar.xz There are .gz and .sig files here, too: http://people.redhat.com/meyering/cu/coreutils-ss.tar.xz http://people.redhat.com/meyering/cu/coreutils-8.1.30-0f8bb.tar.xz Changes in coreutils since 8.1.24-7a2b0: Jim Meyering (1): build: update gnulib submodule to latest Pádraig Brady (5): tests: fix stty failure with serial control settings tests: fix stty-row-col failure on small fixed terminals timeout: fix failure if timeout's parent has ignored SIGCHLD sort: fix failure if sort's parent has ignored SIGCHLD maint: remove an erroneous comment Changes in gnulib since 8.1.24-7a2b0: * gnulib 3e035a5...5aad650 (22): > init.sh: avoid Solaris 10 /bin/sh portability problem > Document pty.h functions. > fchdir: fix logic bugs > dup2: fix logic bugs > unlink: fix m4 detection > unistd-safer: add unit test > cloexec: preserve text vs. binary across dup_cloexec > mgetgroups: reduce duplicate listings > cloexec: Fix possible compilation error. > prognam: Add diagnostic. > pipe2-safer: new module > stdlib-safer: preserve cloexec flag for mkostemp[s] > unistd-safer: allow preservation of cloexec status via flag > test-dup2: enhance test > cloexec: add dup_cloexec > test-xalloc-die: fix test for mingw > test-fseeko: fix test for mingw > autoupdate > progname: Clarify specification. > maint.mk: backslash-escape parens in default regexp > maint.mk: news-check: use grep -E > maintainer-makefile: allow customization of NEWS entry format
Re: git coreutils 'make check' hangs
Ralf Wildenhues writes: > > Hi Pádraig, > > * Pádraig Brady wrote on Tue, Dec 08, 2009 at 02:00:20AM CET: > > Ralf Wildenhues wrote: > > > for some time now, 'make check' in the git coreutils tree hangs for me: > [...] > > I think it may be a gdb issue: > > http://lists.gnu.org/archive/html/bug-coreutils/2009-12/msg00025.htm > > For the moment we've marked that test as "very expensive" > > so that it will not be run by default. > > Ah, ok, I didn't see that thread. Sorry about the noise then. > If yours is caused by the same thing (gdb failing to kill its child process and exit) could you tell me your uname -a? I'd like to be sure it's not just an arch-dependent kernel bug before reporting it as a man page bug. Something still needs to get fixed, even if coreutils no longer cares about it. -- Alan Curry
Re: new snapshot available: coreutils-8.1.30-0f8bb
Passed Skipped Failed \- Fedora core 5 x86 | 363 48 0 Fedora 11 x86 | 362 49 0 Solaris 10 x86| 342 69 0 FreeBSD 6 x86 | 338 73 0
touching "-"
$ touch - && echo ok ok $ ls - ls: cannot access -: No such file or directory $ touch -- - && echo ok ok $ ls - ls: cannot access -: No such file or directory $ : > - $ ls - -
Re: touching "-"
> $ touch - && echo ok > ok Not a bug - you successfully touched stdout (in this case, your terminal). Remember, many commands special case - to mean stdin or stdout if they operate well on a single file. > > $ ls - > ls: cannot access -: No such file or directory ls is not one of those commands. So rather than listing stdin or stdout (neither of which makes sense), it tries to list ./- and fails. If you really must touch ./-, then say so: touch ./- -- Eric Blake
