Bug#219271: libc6 - statvfs broken
Package: libc6 Version: 2.3.2.ds1-9 Tags: sid Hi, The .ds1 versions of libc6 broke something, presumably in statvfs(). I have a remote (stable, 2.4.something) system with an unstable chroot on it. Inside that chroot only a nfs share is mounted, nothing else. Up to last week all worked fine; however since last time I updated the chroot libc6 got upgraded to .ds1-8. When I tried updating the chroot again today apt-get hung where it should prompt me to continue (i.e., after listing what it was about to do). Then I unmounted the nfs share, but to no effect - apt-get still hung. Nothing was mounted (inside the chroot) anymore at that point. I fiddled around a bit, with little success, until I mounted /proc to be able to run ps without having to drop out of the chroot. When I stupidly ran apt-get next, it suddenly worked. Unmounted /proc, hung again. I ran strace and ltrace on an apt-get invocation next. The strace output shows that apt-get opens /etc/mtab, then gets into an infinite loop: ... write(1, "Need to get 8671kB/9020kB of arc"..., 39) = 39 write(1, "After unpacking 26.0MB of additi"..., 62) = 62 statfs("/var/cache/apt/archives/", {f_type="EXT2_SUPER_MAGIC", f_bsize=4096, f_blocks=9376915, f_bfree=3701352, f_bavail=3225026, f_files=4767744, f_ffree=4227550, f_fsid={0, 0}, f_namelen=255, f_frsize=0}) = 0 stat64("/var/cache/apt/archives/", {st_mode=S_IFDIR|0755, st_size=135168, ...}) = 0 open("/proc/mounts", O_RDONLY) = -1 ENOENT (No such file or directory) open("/etc/mtab", O_RDONLY) = 15 fstat64(15, {st_mode=S_IFREG|0644, st_size=0, ...}) = 0 old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40019000 read(15, "", 4096) = 0 _llseek(15, 0, [0], SEEK_SET) = 0 read(15, "", 4096) = 0 _llseek(15, 0, [0], SEEK_SET) = 0 read(15, "", 4096) = 0 _llseek(15, 0, [0], SEEK_SET) = 0 (repeated several thousand times) read(15, "", 4096) = 0 _llseek(15, 0, [0], SEEK_SET) = 0 --- SIGINT (Interrupt) @ 0 (0) --- +++ killed by SIGINT +++ Note that opening /proc/mounts is tried before mtab, which is supposedly why things work when /proc is mounted (altough it's curious that it does not loop then). Note also that currently mtab is empty, but it hardly matters. After mounting the nfs share again, it reads a number of bytes, stat64()s the nfs mount point as it does with /var/cache/apt/archives/ now, does another read() with 0 result as happens above too, and then there's the _llseek() and things start all over again. (Would the _llseek() be the culprit?) Running from within ltrace rather than strace gives this as final output: ... dcgettext(0, 0x08060e60, 5, 0x0806989c, 0xbfffef40)= 0x08060e60 _Z8ioprintfRSoPKcz(0x08065c20, 0x08060e60, 0x0806989c, 0x0806989c, 0xbfffef40) = 0x08065c20 _ZNSs4_Rep10_M_destroyERKSaIcE(0x08069890, 0xb460, 0x0806989c, 0x0806989c, 0xbfffef40) = 0 _Z12_GetErrorObjv(0x08069890, 0xb460, 0x0806989c, 0x0806989c, 0xbfffef40) = 0x08068e40 _ZNK13Configuration5FindBEPKcb(0x080660e0, 0x0805fc30, 0, 0x0806989c, 0xbfffef40) = 0 _ZNK13Configuration5FindBEPKcb(0x080660e0, 0x0805fc4b, 1, 0x0806989c, 0xbfffef40) = 1 _ZNK13Configuration7FindDirEPKcS1_(0xb440, 0x080660e0, 0x0805fd03, 0, 0xbfffef40) = 0xb440 statvfs(0x08069484, 0xbfffef30, 0x0805fd03, 0, 0xbfffef40 --- SIGINT (Interrupt) --- +++ killed by SIGINT +++ Hence my assumption to blame statvfs(). I downgraded to 2.3.2-9 from snapshot.debian.net; this version worked both with and without /proc being mounted. Regards, Filip -- Evil Overlord Quote of the Day: 150.I will provide funding and research to develop tactical and strategic weapons covering a full range of needs so my choices are not limited to "hand to hand combat with swords" and "blow up the planet". -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#174027: libc6 2.3.1-7 still broken
severity 174027 critical thanks (I'm upping the severity, as this definitely fits the description of "making unrelated software break") On Mon, Dec 23, 2002 at 06:58:50PM -0800, Jeff Bailey wrote: > On Mon, Dec 23, 2002 at 08:02:45PM -0600, FireEgl wrote: > > The problem still exists in 2.3.1-7. > > I can't reproduce it - Can you try to debug it a bit? Same here, both with -6 and -7, -5 is fine. PIII 450. Running lynx in gdb, with libc6-dbg -7 installed (and LD_LIBRARY_PATH set): (gdb) run http://bugs.debian.org/libc6 (no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)... Program received signal SIGSEGV, Segmentation fault. 0x in ?? () (gdb) bt #0 0x in ?? () #1 0x401fdfa8 in gaih_inet (name=0x8176138 "bugs.debian.org", service=0xbfffe7bc, req=0xbfffe7c8, pai=0xbfffe7b8) at ../sysdeps/posix/getaddrinfo.c:573 #2 0x401feaed in *__GI_getaddrinfo (name=0x8176138 "bugs.debian.org", service=0x400 , hints=0xbfffe7c8, pai=0xbfffe7b8) at ../sysdeps/posix/getaddrinfo.c:880 #3 0x080d5aff in strcpy () at ../sysdeps/generic/strcpy.c:31 #4 0x080d5c9d in strcpy () at ../sysdeps/generic/strcpy.c:31 #5 0x080c885a in strcpy () at ../sysdeps/generic/strcpy.c:31 #6 0x080c77c3 in strcpy () at ../sysdeps/generic/strcpy.c:31 #7 0x080c7b8d in strcpy () at ../sysdeps/generic/strcpy.c:31 #8 0x080c806e in strcpy () at ../sysdeps/generic/strcpy.c:31 #9 0x08067a55 in strcpy () at ../sysdeps/generic/strcpy.c:31 #10 0x08074419 in strcpy () at ../sysdeps/generic/strcpy.c:31 #11 0x0806a329 in strcpy () at ../sysdeps/generic/strcpy.c:31 #12 0x401629f1 in __libc_start_main (main=0x8068cf8 , argc=2, ubp_av=0xbfffe7e8, init=0x804a710 <_init>, fini=0x80f9450 <_fini>, rtld_fini=0x8176138, stack_end=0x400) at ../sysdeps/generic/libc-start.c:147 (gdb) Apt-get fails, straw fails, I suppose others too but I caught it immediately after upgrading to -6 yesterday so I downgraded and didn't try others. Oh, reportbug just failed, too (I forgot to downgrade again after the above gdb run) Regards, Filip -- "The man who trades freedom for security does not deserve nor will he ever receive either." -- Benjamin Franklin
Bug#174027: libc6 2.3.1-7 still broken
severity 174027 critical thanks (I'm upping the severity, as this definitely fits the description of "making unrelated software break") On Mon, Dec 23, 2002 at 06:58:50PM -0800, Jeff Bailey wrote: > On Mon, Dec 23, 2002 at 08:02:45PM -0600, FireEgl wrote: > > The problem still exists in 2.3.1-7. > > I can't reproduce it - Can you try to debug it a bit? Same here, both with -6 and -7, -5 is fine. PIII 450. Running lynx in gdb, with libc6-dbg -7 installed (and LD_LIBRARY_PATH set): (gdb) run http://bugs.debian.org/libc6 (no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)... Program received signal SIGSEGV, Segmentation fault. 0x in ?? () (gdb) bt #0 0x in ?? () #1 0x401fdfa8 in gaih_inet (name=0x8176138 "bugs.debian.org", service=0xbfffe7bc, req=0xbfffe7c8, pai=0xbfffe7b8) at ../sysdeps/posix/getaddrinfo.c:573 #2 0x401feaed in *__GI_getaddrinfo (name=0x8176138 "bugs.debian.org", service=0x400 , hints=0xbfffe7c8, pai=0xbfffe7b8) at ../sysdeps/posix/getaddrinfo.c:880 #3 0x080d5aff in strcpy () at ../sysdeps/generic/strcpy.c:31 #4 0x080d5c9d in strcpy () at ../sysdeps/generic/strcpy.c:31 #5 0x080c885a in strcpy () at ../sysdeps/generic/strcpy.c:31 #6 0x080c77c3 in strcpy () at ../sysdeps/generic/strcpy.c:31 #7 0x080c7b8d in strcpy () at ../sysdeps/generic/strcpy.c:31 #8 0x080c806e in strcpy () at ../sysdeps/generic/strcpy.c:31 #9 0x08067a55 in strcpy () at ../sysdeps/generic/strcpy.c:31 #10 0x08074419 in strcpy () at ../sysdeps/generic/strcpy.c:31 #11 0x0806a329 in strcpy () at ../sysdeps/generic/strcpy.c:31 #12 0x401629f1 in __libc_start_main (main=0x8068cf8 , argc=2, ubp_av=0xbfffe7e8, init=0x804a710 <_init>, fini=0x80f9450 <_fini>, rtld_fini=0x8176138, stack_end=0x400) at ../sysdeps/generic/libc-start.c:147 (gdb) Apt-get fails, straw fails, I suppose others too but I caught it immediately after upgrading to -6 yesterday so I downgraded and didn't try others. Oh, reportbug just failed, too (I forgot to downgrade again after the above gdb run) Regards, Filip -- "The man who trades freedom for security does not deserve nor will he ever receive either." -- Benjamin Franklin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]