2.6.18-rc1-mm2: process `showconsole' used the removed sysctl system call
After installing a 2.6.18-rc1-mm2 kernel without sysctl syscall support on a standard SuSE 10.0 system, I find the following in my dmesg:
Quoted 5 lines
[ 36.955720] warning: process `showconsole' used the removed sysctl system call [ 39.656410] warning: process `showconsole' used the removed sysctl system call [ 43.304401] warning: process `showconsole' used the removed sysctl system call [ 45.717220] warning: process `ls' used the removed sysctl system call [ 45.789845] warning: process `touch' used the removed sysctl system call
which at face value seems to contradict the statement in the help text for the CONFIG_SYSCTL_SYSCALL option that "Nothing has been using the binary sysctl interface for some time time now". (sic)
Meanwhile, the second part of that sentence that "nothing should break" by disabling it seems to hold true anyway. The system runs fine, and activating CONFIG_SYSCTL_SYSCALL in the kernel doesn't seem to have any effect apart from changing the word "removed" to "obsolete" in the above messages.
HTH
Tilman
Signature · 2 lines
Tilman Schmidt E-Mail: [email protected] Bonn, Germany
Quoted 13 lines · On Sat, 15 Jul 2006 16:40:36 +0200 Tilman Schmidt <[email protected]> wrote:
After installing a 2.6.18-rc1-mm2 kernel without sysctl syscall support on a standard SuSE 10.0 system, I find the following in my dmesg:
Quoted 5 lines
[ 36.955720] warning: process `showconsole' used the removed sysctl system call [ 39.656410] warning: process `showconsole' used the removed sysctl system call [ 43.304401] warning: process `showconsole' used the removed sysctl system call [ 45.717220] warning: process `ls' used the removed sysctl system call [ 45.789845] warning: process `touch' used the removed sysctl system callwhich at face value seems to contradict the statement in the help text for the
CONFIG_SYSCTL_SYSCALLoption that "Nothing has been using the binary sysctl interface for some time time now". (sic)
Quoted 5 lines
Meanwhile, the second part of that sentence that "nothing should break" by disabling it seems to hold true anyway. The system runs fine, and activating
CONFIG_SYSCTL_SYSCALLin the kernel doesn't seem to have any effect apart from changing the word "removed" to "obsolete" in the above messages.
Thanks.
Eric, that tends to make the whole idea inviable, doesn't it?
Quoted 25 lines · Andrew Morton
Quoted 4 lines · On Sat, 15 Jul 2006 16:40:36 +0200 Tilman Schmidt <[email protected]> wrote:
After installing a 2.6.18-rc1-mm2 kernel without sysctl syscall support on a standard SuSE 10.0 system, I find the following in my dmesg:
Quoted 1 line
[ 36.955720] warning: process `showconsole' used the removed sysctl systemcall
Quoted 1 line
Quoted 1 line
[ 39.656410] warning: process `showconsole' used the removed sysctl systemcall
Quoted 1 line
Quoted 1 line
[ 43.304401] warning: process `showconsole' used the removed sysctl systemcall
Quoted 7 lines
Quoted 2 lines
[ 45.717220] warning: process `ls' used the removed sysctl system call [ 45.789845] warning: process `touch' used the removed sysctl system callwhich at face value seems to contradict the statement in the help text for the
CONFIG_SYSCTL_SYSCALLoption that "Nothing has been using the binary sysctl interface for some time time now". (sic)Quoted 5 lines
Meanwhile, the second part of that sentence that "nothing should break" by disabling it seems to hold true anyway. The system runs fine, and activating
CONFIG_SYSCTL_SYSCALLin the kernel doesn't seem to have any effect apart from changing the word "removed" to "obsolete" in the above messages.
Quoted 3 lines
Thanks.
Eric, that tends to make the whole idea inviable, doesn't it?
Close but not quite.
It is the glibc pthread library testing to see if you have an SMP kernel by greping for SMP in the UTS_VERSION string. glibc has always had a fallback to using /proc/sys/kernel/version so it will behave properly, and uname is really the right interface to this data.
My next step in testing is to remove that stupid usage from glibc and see if any other warnings happen. I don't have time to finish setting up that test before I leave for Ottowa.
I have already sent Ulrich Drepper a patch to have glibc just use uname. The patch was posted to both linux-kernel and libc-alpha.
So yes my understanding was not quite correct, but while I have the details wrong I don't know that the substance is wrong. I need to take this at least to the next step so a convincing argument can be made for keeping sys_sysctl.
If sys_sysctl really proves to be used and useful we can say yes people really do use things thing. We really need to support this. We can then revert the 2003 deprecated comment from the header file and drop patches that do not properly maintain the binary interface to sys_sysctl.
I don't really care either way but I want a good case made so sys_sysctl can stop being walking dead. I hate the attitude of have a interface from kernel to user space where people do not care if they break backwards binary compatibility.
Eric
Hi Andrew,
Quoted 15 lines · Andrew Morton
Quoted 12 lines · On Sat, 15 Jul 2006 16:40:36 +0200 Tilman Schmidt <[email protected]> wrote:
After installing a 2.6.18-rc1-mm2 kernel without sysctl syscall support on a standard SuSE 10.0 system, I find the following in my dmesg:
Quoted 5 lines
[ 36.955720] warning: process `showconsole' used the removed sysctl system call [ 39.656410] warning: process `showconsole' used the removed sysctl system call [ 43.304401] warning: process `showconsole' used the removed sysctl system call [ 45.717220] warning: process `ls' used the removed sysctl system call [ 45.789845] warning: process `touch' used the removed sysctl system callwhich at face value seems to contradict the statement in the help text for the
CONFIG_SYSCTL_SYSCALLoption that "Nothing has been using the binary sysctl interface for some time time now". (sic)
Quoted 8 lines
Quoted 5 lines
Meanwhile, the second part of that sentence that "nothing should break" by disabling it seems to hold true anyway. The system runs fine, and activating
CONFIG_SYSCTL_SYSCALLin the kernel doesn't seem to have any effect apart from changing the word "removed" to "obsolete" in the above messages.Thanks.
date and salsa also use sysctl.
warning: process `date' used the removed sysctl system call
warning: process `salsa' used the removed sysctl system call
Quoted 1 line
Eric, that tends to make the whole idea inviable, doesn't it?
How about _very_ long term to remove sysctl (i.e. January 2010)?
Regards,
Michal
Signature · 25 lines
Michal K. K. Piotrowski LTG - Linux Testers Group (http://www.stardust.webpages.pl/ltg/wiki/) Signed-off-by: Michal Piotrowski <[email protected]> diff -uprN -X linux-mm/Documentation/dontdiff linux-mm-clean/Documentation/ABI/obsolete/sysctl linux-mm/Documentation/ABI/obsolete/sysctl --- linux-mm-clean/Documentation/ABI/obsolete/sysctl 1970-01-01 01:00:00.000000000 +0100 +++ linux-mm/Documentation/ABI/obsolete/sysctl 2006-07-16 01:22:51.000000000 +0200 @@ -0,0 +1,15 @@ +What: sys_sysctl +Date: January 2010 +Contact: Eric W. Biederman <[email protected]> +Description: + sys_sysctl uses binary paths that have been found to be a major + pain to maintain and use. The interface in /proc/sys is now + the primary and what everyone uses. + + Nothing has been using the binary sysctl interface for some time + time now so nothing should break if you disable sysctl syscall + support, and you kernel will get marginally smaller. + +Users: + +date, ls, salsa, showconsole, touch
Quoted 30 lines · Michal Piotrowski
Hi Andrew,
Quoted 7 lines · Andrew Morton
Quoted 4 lines · On Sat, 15 Jul 2006 16:40:36 +0200 Tilman Schmidt <[email protected]> wrote:
After installing a 2.6.18-rc1-mm2 kernel without sysctl syscall support on a standard SuSE 10.0 system, I find the following in my dmesg:
Quoted 1 line
[ 36.955720] warning: process `showconsole' used the removed sysctl systemcall
Quoted 1 line
Quoted 1 line
Quoted 1 line
[ 39.656410] warning: process `showconsole' used the removed sysctl systemcall
Quoted 1 line
Quoted 1 line
Quoted 1 line
[ 43.304401] warning: process `showconsole' used the removed sysctl systemcall
Quoted 6 lines
Quoted 6 lines
Quoted 2 lines
[ 45.717220] warning: process `ls' used the removed sysctl system call [ 45.789845] warning: process `touch' used the removed sysctl system callwhich at face value seems to contradict the statement in the help text for the
CONFIG_SYSCTL_SYSCALLoption that "Nothing has been using the binary sysctl interface for some time time now". (sic)Quoted 8 lines
Quoted 5 lines
Meanwhile, the second part of that sentence that "nothing should break" by disabling it seems to hold true anyway. The system runs fine, and activating
CONFIG_SYSCTL_SYSCALLin the kernel doesn't seem to have any effect apart from changing the word "removed" to "obsolete" in the above messages.Thanks.
Quoted 8 lines
date and salsa also use sysctl.
warning: process `date' used the removed sysctl system call
warning: process `salsa' used the removed sysctl system callQuoted 1 line
Eric, that tends to make the whole idea inviable, doesn't it?
How about _very_ long term to remove sysctl (i.e. January 2010)?
That may be reasonable. However please confirm that everything that you have complaints from is using libpthreads.
As there is one use of libpthreads that is using sysctl in a very non-serious way.
With libptrheads modified to use uname and not sysctl I am not seeing that message. I thought I had broken my test setup by forgetting to compile glibc with --with-tls but I managed but I managed to get things working again using LD_ASSUME_KERNEL=2.4.1
Still not the best data point but a very interesting one.
Eric
Quoted 37 lines · Eric W. Biederman
Quoted 35 lines · Michal Piotrowski
Hi Andrew,
Quoted 7 lines · Andrew Morton
Quoted 4 lines · On Sat, 15 Jul 2006 16:40:36 +0200 Tilman Schmidt <[email protected]> wrote:
After installing a 2.6.18-rc1-mm2 kernel without sysctl syscall support on a standard SuSE 10.0 system, I find the following in my dmesg:
Quoted 1 line
[ 36.955720] warning: process `showconsole' used the removed sysctl systemcall
Quoted 1 line
Quoted 1 line
Quoted 1 line
[ 39.656410] warning: process `showconsole' used the removed sysctl systemcall
Quoted 1 line
Quoted 1 line
Quoted 1 line
[ 43.304401] warning: process `showconsole' used the removed sysctl systemcall
Quoted 6 lines
Quoted 6 lines
Quoted 2 lines
[ 45.717220] warning: process `ls' used the removed sysctl system call [ 45.789845] warning: process `touch' used the removed sysctl system callwhich at face value seems to contradict the statement in the help text for the
CONFIG_SYSCTL_SYSCALLoption that "Nothing has been using the binary sysctl interface for some time time now". (sic)Quoted 8 lines
Quoted 5 lines
Meanwhile, the second part of that sentence that "nothing should break" by disabling it seems to hold true anyway. The system runs fine, and activating
CONFIG_SYSCTL_SYSCALLin the kernel doesn't seem to have any effect apart from changing the word "removed" to "obsolete" in the above messages.Thanks.
date and salsa also use sysctl.
warning: process `date' used the removed sysctl system call
warning: process `salsa' used the removed sysctl system call
Quoted 6 lines
Quoted 3 lines
Quoted 1 line
Eric, that tends to make the whole idea inviable, doesn't it?
How about _very_ long term to remove sysctl (i.e. January 2010)?
That may be reasonable. However please confirm that everything that you have complaints from is using libpthreads.
ldd /bin/date
linux-gate.so.1 => (0xb7f22000)
librt.so.1 => /lib/librt.so.1 (0x4aab9000)
libc.so.6 => /lib/libc.so.6 (0x49ca3000)
libpthread.so.0 => /lib/libpthread.so.0 (0x49f38000)
/lib/ld-linux.so.2 (0x49c86000)ldd /bin/ls
linux-gate.so.1 => (0xb7f23000)
librt.so.1 => /lib/librt.so.1 (0x4aab9000)
libselinux.so.1 => /lib/libselinux.so.1 (0x45a71000)
libc.so.6 => /lib/libc.so.6 (0x49ca3000)
libpthread.so.0 => /lib/libpthread.so.0 (0x49f38000)
/lib/ld-linux.so.2 (0x49c86000)
libdl.so.2 => /lib/libdl.so.2 (0x49dd8000)
libsepol.so.1 => /lib/libsepol.so.1 (0x41ad5000)ldd /sbin/salsa
linux-gate.so.1 => (0xb7fd7000)
libasound.so.2 => /lib/libasound.so.2 (0x4100f000)
libc.so.6 => /lib/libc.so.6 (0x49ca3000)
libm.so.6 => /lib/libm.so.6 (0x49dde000)
libdl.so.2 => /lib/libdl.so.2 (0x49dd8000)
libpthread.so.0 => /lib/libpthread.so.0 (0x49f38000)
/lib/ld-linux.so.2 (0x49c86000)ldd /bin/touch
linux-gate.so.1 => (0xb7fa2000)
librt.so.1 => /lib/librt.so.1 (0x4aab9000)
libc.so.6 => /lib/libc.so.6 (0x49ca3000)
libpthread.so.0 => /lib/libpthread.so.0 (0x49f38000)
/lib/ld-linux.so.2 (0x49c86000)I can confirm this, but I don't have a "showconsole".
Quoted 4 lines
As there is one use of libpthreads that is using sysctl in a very non-serious way.
Quoted 9 lines
With libptrheads modified to use uname and not sysctl I am not seeing that message. I thought I had broken my test setup by forgetting to compile glibc with --with-tls but I managed but I managed to get things working again using LD_ASSUME_KERNEL=2.4.1
Still not the best data point but a very interesting one.
Eric
Regards,
Michal
Signature · 3 lines
Michal K. K. Piotrowski LTG - Linux Testers Group (http://www.stardust.webpages.pl/ltg/wiki/)