Bug#403332: util-vserver can not hashify a guest
tags 403332 + patch thanks On Sun, Feb 11, 2007 at 01:27:01PM +0100, Daniel Dehennin wrote: Le 4878 Septembre 1993, Ola Lundqvist a tapoté: Ohh. Thanks a lot then! The problem is resolved by the syscall_shiny14e.h from http://vserver.13thfloor.at/Experimental/SYSCALL/syscall_shiny14e.h With this, it works with debian GCC 4.1. Thanks. I'll apply this when etch is out of the door. Regards, // Ola Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- - Ola Lundqvist --- / [EMAIL PROTECTED] Annebergsslingan 37 \ | [EMAIL PROTECTED] 654 65 KARLSTAD | | +46 (0)54-10 14 30 +46 (0)70-332 1551 | | http://opalsys.net/ UIN/icq: 4912500 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Le 4878 Septembre 1993, Ola Lundqvist a tapoté: Ohh. Thanks a lot then! The problem is resolved by the syscall_shiny14e.h from http://vserver.13thfloor.at/Experimental/SYSCALL/syscall_shiny14e.h With this, it works with debian GCC 4.1. Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi Daniel Thanks a lot for your help. I have a question about the other part of the patch, and that is the removal of the useless syscall things thing. Is that needed for this to work? The other question is if a condition about gcc = 4.1 was needed or if gcc-4.1 was enough? Regards, // Ola On Mon, Jan 08, 2007 at 07:48:52PM +0100, Daniel Dehennin wrote: Le 4859 Septembre 1993, Ola Lundqvist a tapoté: Hi again Hello, Now I know that gcc 4.2 is not generally available unfortunatly so that is not an option... I tried with experimental gcc-4.2 and the problem persists. I wonder which alternative that is best. To disable optimisation or remove --enable-apis=NOLEGACY... For sparc only of course. According to Daniel Hozac, the only workaround he can see is --disable-alternative-syscalls, I test it and it works. I build a debian package and it works, I include a diff. Maybe adding a condition about gcc = 4.1 ? I remove unnecessary definition of syscall when using dietlibc with traditional syscalls. Regards. -- Daniel Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x6A2540D1 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Le 4859 Septembre 1993, Ola Lundqvist a tapoté:
Hi again
Hello,
Now I know that gcc 4.2 is not generally available unfortunatly
so that is not an option...
I tried with experimental gcc-4.2 and the problem persists.
I wonder which alternative that is best. To disable optimisation or
remove --enable-apis=NOLEGACY... For sparc only of course.
According to Daniel Hozac, the only workaround he can see is
--disable-alternative-syscalls, I test it and it works.
I build a debian package and it works, I include a diff.
Maybe adding a condition about gcc = 4.1 ?
I remove unnecessary definition of syscall when using dietlibc with
traditional syscalls.
Regards.
--
Daniel Dehennin
Récupérer ma clef GPG:
gpg --keyserver pgp.mit.edu --recv-keys 0x6A2540D1
pgpwUOE8PtHzw.pgp
Description: PGP signature
diff -Nru util-vserver-0.30.212.old/debian/changelog util-vserver-0.30.212/debian/changelog
--- util-vserver-0.30.212.old/debian/changelog 2007-01-08 18:09:28.0 +
+++ util-vserver-0.30.212/debian/changelog 2007-01-08 17:47:39.0 +
@@ -1,3 +1,11 @@
+util-vserver (0.30.212-1-20070108) unstable; urgency=low
+
+ * Use --disable-alternative-syscalls on sparc
+ * dietlibc already declare syscall, remove useless ifdef __dietlibc__ with
+ENSC_SYSCALL_TRADITIONAL
+
+ -- Daniel Dehennin [EMAIL PROTECTED] Mon, 8 Jan 2007 18:44:10 +0100
+
util-vserver (0.30.212-1) unstable; urgency=low
* New upstream release
diff -Nru util-vserver-0.30.212.old/debian/rules util-vserver-0.30.212/debian/rules
--- util-vserver-0.30.212.old/debian/rules 2007-01-08 18:09:28.0 +
+++ util-vserver-0.30.212/debian/rules 2007-01-08 17:40:19.0 +
@@ -40,6 +40,7 @@
# says this works and it is anyway the oldest sparc generation we
# support in Debian (or kernel/glibc).
CFLAGS += -mcpu=v8
+confflags += --disable-alternative-syscalls
endif
ifeq ($(DEB_HOST_ARCH),hppa)
diff -Nru util-vserver-0.30.212.old/lib/vserver-internal.h util-vserver-0.30.212/lib/vserver-internal.h
--- util-vserver-0.30.212.old/lib/vserver-internal.h 2006-12-09 16:15:31.0 +
+++ util-vserver-0.30.212/lib/vserver-internal.h 2007-01-08 17:45:08.0 +
@@ -253,10 +253,6 @@
inline static UNUSED ALWAYSINLINE
int vserver(uint32_t cmd, uint32_t id, void *data)
{
-#if defined __dietlibc__
- extern long int syscall (long int __sysno, ...);
-#endif
-
return syscall(__NR_vserver, cmd, id, data);
}
#else
diff -Nru util-vserver-0.30.212.old/lib_internal/sys_clone.h util-vserver-0.30.212/lib_internal/sys_clone.h
--- util-vserver-0.30.212.old/lib_internal/sys_clone.h 2006-12-09 16:15:35.0 +
+++ util-vserver-0.30.212/lib_internal/sys_clone.h 2007-01-08 17:52:26.0 +
@@ -42,9 +42,6 @@
#ifdef __sparc__
int parent = getpid();
#endif
-#if defined(__dietlibc__) defined(ENSC_SYSCALL_TRADITIONAL)
- extern long int syscall (long int __sysno, ...);
-#endif
#if defined(__s390__) defined(ENSC_SYSCALL_TRADITIONAL)
ret = syscall(__NR__sys_clone, child_stack, flags);
diff -Nru util-vserver-0.30.212.old/lib_internal/sys_personality.h util-vserver-0.30.212/lib_internal/sys_personality.h
--- util-vserver-0.30.212.old/lib_internal/sys_personality.h 2006-12-09 16:15:35.0 +
+++ util-vserver-0.30.212/lib_internal/sys_personality.h 2007-01-08 17:56:37.0 +
@@ -28,10 +28,6 @@
inline static UNUSED ALWAYSINLINE
int sys_personality(int pers)
{
-#if defined __dietlibc__
- extern long int syscall (long int __sysno, ...);
-#endif
-
return syscall(__NR_sys_personality, pers);
}
#else
Bug#403332: util-vserver can not hashify a guest
Le 4877 Septembre 1993, Ola Lundqvist a tapoté: I have a question about the other part of the patch, and that is the removal of the useless syscall things thing. Is that needed for this to work? Yes, if you disable the alternative syscall with the use of dietlibc, gcc complains that syscall is already defined in /usr/lib/include/sys/syscall.h. The other question is if a condition about gcc = 4.1 was needed or if gcc-4.1 was enough? Well, I made a test with gcc 4.2 and it doesn't work... -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Le 4877 Septembre 1993, Ola Lundqvist a tapoté: Now I'm confused. What did not work with 4.2? Sorry, the unmodified debian source package does not work with gcc-4.2. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Mon, Jan 08, 2007 at 08:39:58PM +0100, Daniel Dehennin wrote: Le 4877 Septembre 1993, Ola Lundqvist a tapoté: I have a question about the other part of the patch, and that is the removal of the useless syscall things thing. Is that needed for this to work? Yes, if you disable the alternative syscall with the use of dietlibc, gcc complains that syscall is already defined in /usr/lib/include/sys/syscall.h. The other question is if a condition about gcc = 4.1 was needed or if gcc-4.1 was enough? Well, I made a test with gcc 4.2 and it doesn't work... Now I'm confused. What did not work with 4.2? Regards, // Ola -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Mon, Jan 08, 2007 at 09:16:21PM +0100, Daniel Dehennin wrote: Le 4877 Septembre 1993, Ola Lundqvist a tapoté: Now I'm confused. What did not work with 4.2? Sorry, the unmodified debian source package does not work with gcc-4.2. I see. Thanks. Would it be possible for you to try to compile it with 4.1 as well? If that works I can upload this correction to unstable at once. I may even be able to get that through to frozen/testing/etch but that is not that sure. Regards, // Ola -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Le 4877 Septembre 1993, Ola Lundqvist a tapoté: I see. Thanks. Would it be possible for you to try to compile it with 4.1 as well? If that works I can upload this correction to unstable at once. Well, my package is build with gcc-4.1, the 4.2 test was to see if things work out of the box with this version of gcc. Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Tue, Jan 09, 2007 at 07:31:33AM +0100, Daniel Dehennin wrote: Le 4877 Septembre 1993, Ola Lundqvist a tapoté: I see. Thanks. Would it be possible for you to try to compile it with 4.1 as well? If that works I can upload this correction to unstable at once. Well, my package is build with gcc-4.1, the 4.2 test was to see if things work out of the box with this version of gcc. Ohh. Thanks a lot then! Regards, // Ola Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Sat, Dec 23, 2006 at 12:43:45AM +0100, Daniel Hokka Zakrisson wrote: Ola Lundqvist wrote: Hi On Fri, Dec 22, 2006 at 04:16:06PM +0100, Daniel Hokka Zakrisson wrote: Ola Lundqvist wrote: Hi ...CUT... Unfortunately I have no idea how to use the Debian BTS (and I'm not a Debian user nor developer, so it would be somewhat strange if the report came from me). If you give me a bug number, I'd be more than happy to add my analysis to it though. (Or I could add it here, if you think it's relevant.) Oh I was not aware of that. Please add the analysis to this bug (cc: [EMAIL PROTECTED]) so that I can make a clone for this bug to the gcc maintainers. Humm. So now that I was writing my reply, I can't seem to find anything really seriously wrong with the Debian binary, other than it failing to store the return value from the syscall: I see. http://www.asgardr.info/~nebu/vserver/vc_get_iattr_debian_sid.asm: 6a4c:91 d0 20 10 ta 0x10 6a50:a2 40 00 00 addx %g0, %g0, %l1 6a54:80 a4 60 00 cmp %l1, 0 6a58:12 80 00 66 bne 6bf0 vc_get_iattr+0x22c 6a5c:01 00 00 00 nop ... 6bf0:40 00 50 64 call 1ad80 [EMAIL PROTECTED] 6bf4:01 00 00 00 nop 6bf8:d0 22 00 00 st %o0, [ %o0 ] But this shouldn't cause the issue in this bug. I'm sorry for my premature comments, we'll have to do another round of debugging to find the exact cause... Ok I see. Well if you can determine this it is really good. For me it is enough to know that gcc-4.2 solves it. However with the knowledge that you may discover it may be possible to solve it in util-vserver instead. Regards, // Ola -- Daniel Hokka Zakrisson GPG id: 06723412 GPG fingerprint: A455 4DF3 990A 431F FECA 7947 6136 DDA2 0672 3412 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Fri, Dec 22, 2006 at 12:17:52AM +0100, Daniel Hokka Zakrisson wrote: Ola Lundqvist wrote: Hi again Now I know that gcc 4.2 is not generally available unfortunatly so that is not an option... I wonder which alternative that is best. To disable optimisation or remove --enable-apis=NOLEGACY... For sparc only of course. I don't think the latter is sufficient. If vc_get_iattr is any indication, the current compiler's optimizations are _very_ broken. A lot of the instructions are ordered incorrectly. I think disabling the optimizations is the only way to fix it locally, though I guess gcc really ought to be fixed. Maybe this bug should be forwarded to the gcc package? Or can you write a new bug against that package and mark this one as related? I can do that of course but you seem to know in more details what is wrong. Regards, // Ola -- Daniel Hokka Zakrisson GPG id: 06723412 GPG fingerprint: A455 4DF3 990A 431F FECA 7947 6136 DDA2 0672 3412 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Ola Lundqvist wrote: Hi On Fri, Dec 22, 2006 at 12:17:52AM +0100, Daniel Hokka Zakrisson wrote: Ola Lundqvist wrote: Hi again Now I know that gcc 4.2 is not generally available unfortunatly so that is not an option... I wonder which alternative that is best. To disable optimisation or remove --enable-apis=NOLEGACY... For sparc only of course. I don't think the latter is sufficient. If vc_get_iattr is any indication, the current compiler's optimizations are _very_ broken. A lot of the instructions are ordered incorrectly. I think disabling the optimizations is the only way to fix it locally, though I guess gcc really ought to be fixed. Maybe this bug should be forwarded to the gcc package? Or can you write a new bug against that package and mark this one as related? I can do that of course but you seem to know in more details what is wrong. Unfortunately I have no idea how to use the Debian BTS (and I'm not a Debian user nor developer, so it would be somewhat strange if the report came from me). If you give me a bug number, I'd be more than happy to add my analysis to it though. (Or I could add it here, if you think it's relevant.) -- Daniel Hokka Zakrisson GPG id: 06723412 GPG fingerprint: A455 4DF3 990A 431F FECA 7947 6136 DDA2 0672 3412 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Fri, Dec 22, 2006 at 04:16:06PM +0100, Daniel Hokka Zakrisson wrote: Ola Lundqvist wrote: Hi ...CUT... Unfortunately I have no idea how to use the Debian BTS (and I'm not a Debian user nor developer, so it would be somewhat strange if the report came from me). If you give me a bug number, I'd be more than happy to add my analysis to it though. (Or I could add it here, if you think it's relevant.) Oh I was not aware of that. Please add the analysis to this bug (cc: [EMAIL PROTECTED]) so that I can make a clone for this bug to the gcc maintainers. Best regards, // Ola -- Daniel Hokka Zakrisson GPG id: 06723412 GPG fingerprint: A455 4DF3 990A 431F FECA 7947 6136 DDA2 0672 3412 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Ola Lundqvist wrote: Hi On Fri, Dec 22, 2006 at 04:16:06PM +0100, Daniel Hokka Zakrisson wrote: Ola Lundqvist wrote: Hi ...CUT... Unfortunately I have no idea how to use the Debian BTS (and I'm not a Debian user nor developer, so it would be somewhat strange if the report came from me). If you give me a bug number, I'd be more than happy to add my analysis to it though. (Or I could add it here, if you think it's relevant.) Oh I was not aware of that. Please add the analysis to this bug (cc: [EMAIL PROTECTED]) so that I can make a clone for this bug to the gcc maintainers. Humm. So now that I was writing my reply, I can't seem to find anything really seriously wrong with the Debian binary, other than it failing to store the return value from the syscall: http://www.asgardr.info/~nebu/vserver/vc_get_iattr_debian_sid.asm: 6a4c: 91 d0 20 10 ta 0x10 6a50: a2 40 00 00 addx %g0, %g0, %l1 6a54: 80 a4 60 00 cmp %l1, 0 6a58: 12 80 00 66 bne 6bf0 vc_get_iattr+0x22c 6a5c: 01 00 00 00 nop ... 6bf0: 40 00 50 64 call 1ad80 [EMAIL PROTECTED] 6bf4: 01 00 00 00 nop 6bf8: d0 22 00 00 st %o0, [ %o0 ] But this shouldn't cause the issue in this bug. I'm sorry for my premature comments, we'll have to do another round of debugging to find the exact cause... -- Daniel Hokka Zakrisson GPG id: 06723412 GPG fingerprint: A455 4DF3 990A 431F FECA 7947 6136 DDA2 0672 3412 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Le 4856 Septembre 1993, Ola Lundqvist a tapoté: Hello Hello, Can it be the --enable-apis=NOLEGACY ? The problem seems to come from the --enable-apis=NOLEGACY, without it works. #vservers people try to figure out what's the problem, finally I compile the mainline and the debian util-vserver with GCC 3.4 with and without --enable-apis=NOLEGACY and everything works. Then I compile util-vserver with GCC 4.1 without the optimisation -O2 and it works. So, the problem is a miscompilation by GCC 4.1 with -O2. Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi Really good to know. Then it is a problem for sparc with gcc 4.1. I'll see if I can get this kind of update into etch. Regards, // Ola On Thu, Dec 21, 2006 at 05:27:52PM +0100, Daniel Dehennin wrote: Le 4856 Septembre 1993, Ola Lundqvist a tapoté: Hello Hello, Can it be the --enable-apis=NOLEGACY ? The problem seems to come from the --enable-apis=NOLEGACY, without it works. #vservers people try to figure out what's the problem, finally I compile the mainline and the debian util-vserver with GCC 3.4 with and without --enable-apis=NOLEGACY and everything works. Then I compile util-vserver with GCC 4.1 without the optimisation -O2 and it works. So, the problem is a miscompilation by GCC 4.1 with -O2. Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- - Ola Lundqvist --- / [EMAIL PROTECTED] Annebergsslingan 37 \ | [EMAIL PROTECTED] 654 65 KARLSTAD | | +46 (0)54-10 14 30 +46 (0)70-332 1551 | | http://opalsys.net/ UIN/icq: 4912500 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi again Now I know that gcc 4.2 is not generally available unfortunatly so that is not an option... I wonder which alternative that is best. To disable optimisation or remove --enable-apis=NOLEGACY... For sparc only of course. In unstable this will be solved when gcc-4.2 enters the archives but for upcoming stable? Micah, do you have an opinion? Do you think this is severe enough to try to push for etch? Regards, // Ola On Thu, Dec 21, 2006 at 05:27:52PM +0100, Daniel Dehennin wrote: Le 4856 Septembre 1993, Ola Lundqvist a tapoté: Hello Hello, Can it be the --enable-apis=NOLEGACY ? The problem seems to come from the --enable-apis=NOLEGACY, without it works. #vservers people try to figure out what's the problem, finally I compile the mainline and the debian util-vserver with GCC 3.4 with and without --enable-apis=NOLEGACY and everything works. Then I compile util-vserver with GCC 4.1 without the optimisation -O2 and it works. So, the problem is a miscompilation by GCC 4.1 with -O2. Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- - Ola Lundqvist --- / [EMAIL PROTECTED] Annebergsslingan 37 \ | [EMAIL PROTECTED] 654 65 KARLSTAD | | +46 (0)54-10 14 30 +46 (0)70-332 1551 | | http://opalsys.net/ UIN/icq: 4912500 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Ola Lundqvist wrote: Hi again Now I know that gcc 4.2 is not generally available unfortunatly so that is not an option... I wonder which alternative that is best. To disable optimisation or remove --enable-apis=NOLEGACY... For sparc only of course. I don't think the latter is sufficient. If vc_get_iattr is any indication, the current compiler's optimizations are _very_ broken. A lot of the instructions are ordered incorrectly. I think disabling the optimizations is the only way to fix it locally, though I guess gcc really ought to be fixed. -- Daniel Hokka Zakrisson GPG id: 06723412 GPG fingerprint: A455 4DF3 990A 431F FECA 7947 6136 DDA2 0672 3412 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Le 4855 Septembre 1993, Ola Lundqvist a tapoté:
Hi
Hello,
So I think you have a kernel problem here.
I made some test with mainline util-vserver and it works.
I compile util-vserver 0.30.212 with debian dietlibc and debian
libeecrypt and the vhashify works, maybe a 32/64 bits problem ?
The vserver-info output:
Versions:
Kernel: 2.6.19.1-vs2.3.0.4+ancalagon.1
VS-API: 0x00020102
util-vserver: 0.30.212; Dec 18 2006, 18:27:55
Features:
CC: gcc, gcc-4.1 (GCC) 4.1.2 20061115 (prerelease)
(Debian 4.1.1-21)
CXX: g++, g++-4.1 (GCC) 4.1.2 20061115 (prerelease)
(Debian 4.1.1-21)
CPPFLAGS: ''
CFLAGS: '-g -O2 -std=c99 -Wall -pedantic -W -funit-at-a-time'
CXXFLAGS: '-g -O2 -ansi -Wall -pedantic -W -fmessage-length=0
-funit-at-a-time'
build/host: sparc64-unknown-linux-gnu/sparc64-unknown-linux-gnu
Use dietlibc: yes
Build C++ programs: yes
Build C99 programs: yes
Available APIs: v13,net,v21
ext2fs Source: kernel
syscall(2) invocation: alternative
vserver(2) syscall#: 267/fallback
Paths:
prefix: /usr/local
sysconf-Directory: ${prefix}/etc
cfg-Directory: ${prefix}/etc/vservers
initrd-Directory: $(sysconfdir)/init.d
pkgstate-Directory: ${prefix}/var/run/vservers
vserver-Rootdir: /vservers
Regards.
--
Daniel 'NebuchadnezzaR' Dehennin
Récupérer ma clef GPG:
gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hello
On Mon, Dec 18, 2006 at 06:57:46PM +0100, Daniel Dehennin wrote:
Le 4855 Septembre 1993, Ola Lundqvist a tapoté:
Hi
Hello,
So I think you have a kernel problem here.
I made some test with mainline util-vserver and it works.
Interesting! The only patch that is applied to the upstream code
is for a configuration directory (scripts/vserver-build.functions)
so it should not be related.
I compile util-vserver 0.30.212 with debian dietlibc and debian
libeecrypt and the vhashify works, maybe a 32/64 bits problem ?
Maybe. How did your --configure line look like?
The debian configure line look like this:
CFLAGS += -mcpu=v8
confflags += --build $(DEB_BUILD_GNU_TYPE) --host $(DEB_HOST_GNU_TYPE)
CFLAGS=$(CFLAGS) ./configure $(confflags) \
--prefix=/usr --enable-release \
--mandir=\$${prefix}/share/man \
--infodir=\$${prefix}/share/info \
--sysconfdir=/etc --enable-dietlibc \
--localstatedir=/var \
--enable-apis=NOLEGACY \
--with-vrootdir=/var/lib/vservers
Can it be the --enable-apis=NOLEGACY ?
Or is it --with-vrootdir maybe?
Regards,
// Ola
The vserver-info output:
Versions:
Kernel: 2.6.19.1-vs2.3.0.4+ancalagon.1
VS-API: 0x00020102
util-vserver: 0.30.212; Dec 18 2006, 18:27:55
Features:
CC: gcc, gcc-4.1 (GCC) 4.1.2 20061115 (prerelease)
(Debian 4.1.1-21)
CXX: g++, g++-4.1 (GCC) 4.1.2 20061115 (prerelease)
(Debian 4.1.1-21)
CPPFLAGS: ''
CFLAGS: '-g -O2 -std=c99 -Wall -pedantic -W
-funit-at-a-time'
CXXFLAGS: '-g -O2 -ansi -Wall -pedantic -W
-fmessage-length=0 -funit-at-a-time'
build/host: sparc64-unknown-linux-gnu/sparc64-unknown-linux-gnu
Use dietlibc: yes
Build C++ programs: yes
Build C99 programs: yes
Available APIs: v13,net,v21
ext2fs Source: kernel
syscall(2) invocation: alternative
vserver(2) syscall#: 267/fallback
Paths:
prefix: /usr/local
sysconf-Directory: ${prefix}/etc
cfg-Directory: ${prefix}/etc/vservers
initrd-Directory: $(sysconfdir)/init.d
pkgstate-Directory: ${prefix}/var/run/vservers
vserver-Rootdir: /vservers
Regards.
--
Daniel 'NebuchadnezzaR' Dehennin
Récupérer ma clef GPG:
gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69
--
--- Ola Lundqvist systemkonsult --- M Sc in IT Engineering
/ [EMAIL PROTECTED] Annebergsslingan 37\
| [EMAIL PROTECTED] 654 65 KARLSTAD|
| http://opalsys.net/ Mobile: +46 (0)70-332 1551 |
\ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 /
---
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi On Sun, Dec 17, 2006 at 01:22:15PM +0100, Daniel Dehennin wrote: Hello, I attach the verbose output of vhashify, it seems that everything is skipped. Seems so yes. Can it be related to that we use beecrypt6 instead of beecrypt2...? Probably not but maybe. Can you increase the verbosity and send me the output? vhashify -v -v -v -v -v -v (-v 6 times). With that you can see the reason for skipping for each file. Regards, // Ola Regards. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 Content-Description: Verbose output of vhashify Initializing exclude-list for /var/lib/vservers/betch (betch) Starting to traverse directories... skipping '/lib/libdl-2.3.6.so' skipping '/lib/ld-2.3.6.so' skipping '/lib/libpamc.so.0.79' skipping '/lib/init/mount-functions.sh' ... skipping '/usr/share/bug/apt/script' skipping '/usr/share/bug/logrotate/script' -- - Ola Lundqvist --- / [EMAIL PROTECTED] Annebergsslingan 37 \ | [EMAIL PROTECTED] 654 65 KARLSTAD | | +46 (0)54-10 14 30 +46 (0)70-332 1551 | | http://opalsys.net/ UIN/icq: 4912500 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
Hi Most of them have reason 'operation not supported'. It is the following vserver kernel syscall that return this information: /** \brief Returns information about attributes and assigned context of a file. * \ingroup syscalls * * This function returns the VC_IATTR_XXX flags and about the assigned * context of a file. To request an information, the appropriate bit in * \c mask must be set and the corresponding parameter (\a xid or \a * flags) must not be NULL. * * E.g. to receive the assigned context, the \c VC_IATTR_XID bit must be * set in \a mask, and \a xid must point to valid memory. * * Possible flags are \c VC_IATTR_ADMIN, \c VC_IATTR_WATCH , \c VC_IATTR_HIDE, * \c VC_IATTR_BARRIER, \c VC_IATTR_IUNLINK and \c VC_IATTR_IMMUTABLE. * * \param filename The name of the file whose attributes shall be determined. * \param xid When non-zero and the VC_IATTR_XID bit is set in \a mask, * the assigned context of \a filename will be stored there. * \param flags When non-zero, a bitmask of current attributes will be * stored there. These attributes must be requested explicitly * by setting the appropriate bit in \a mask * \param mask Points to a bitmask which tells which attributes shall be * determined. On return, it will masquerade the attributes * which were determined. * * \pre mask!=0 !((*maskVC_IATTR_XID) xid==0) !((*mask~VC_IATTR_XID) flags==0) */ int vc_get_iattr(char const *filename, xid_t * /[EMAIL PROTECTED]@*/ xid, uint_least32_t * /[EMAIL PROTECTED]@*/ flags, uint_least32_t * /[EMAIL PROTECTED]@*/ mask) VC_ATTR_NONNULL((1)); According to the code it either: * Return -1 * Stored mask is != flags requested. The requested flags are VC_IATTR_IUNLINK and VC_IATTR_IMMUTABLE. So I think you have a kernel problem here. Regards, // Ola On Sun, Dec 17, 2006 at 07:09:51PM +0100, Daniel Dehennin wrote: Le 4855 Septembre 1993, Ola Lundqvist a tapoté: Can it be related to that we use beecrypt6 instead of beecrypt2...? Probably not but maybe. I don't know. Can you increase the verbosity and send me the output? vhashify -v -v -v -v -v -v (-v 6 times). With that you can see the reason for skipping for each file. Ok, see attached file. -- Daniel 'NebuchadnezzaR' Dehennin Récupérer ma clef GPG: gpg --keyserver pgp.mit.edu --recv-keys 0x2A408F69 -- --- Ola Lundqvist systemkonsult --- M Sc in IT Engineering / [EMAIL PROTECTED] Annebergsslingan 37\ | [EMAIL PROTECTED] 654 65 KARLSTAD| | http://opalsys.net/ Mobile: +46 (0)70-332 1551 | \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / --- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#403332: util-vserver can not hashify a guest
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Daniel Dehennin wrote: Package: util-vserver Version: 0.30.212-1 Severity: normal ... Versions of packages util-vserver depends on: ii debconf 1.5.10 Debian configuration management sy ii iproute 20061002-3 Professional tools to control the ii libbeecrypt6 4.1.2-6 open source C library of cryptogra ii libc62.3.6.ds1-8 GNU C Library: Shared libraries ii make 3.81-3 The GNU version of the make util ii net-tools1.60-17 The NET-3 networking toolkit Versions of packages util-vserver recommends: ii binutils 2.17-3 The GNU assembler, linker and bina ii debootstrap 0.3.3.1Bootstrap a basic Debian system What kernel are you using, with which vserver patch? Micah -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFFhcwe9n4qXRzy1ioRArR4AKCeoRDTwVyt6QgqmzQR3nKwOoXeLwCff2DN 6JqAdZ73U24BxaSDBaT3hrs= =vZ0w -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
