Bug#308792: After the last update Via C3 systems give assertion failed in ld.so at boot

2005-05-12 Thread Anton Ivanov
Package: libc6
Version: 2.3.2.ds1-21
Severity: critical



After the last update C3 Version 1

with a kernel 2.6 image will fail on boot with:

Inconsistency detected by ld.so: do_rel.h: 109 elf_dynamic_do_rel: Assertion 
'(map-l_info[(34+0+(0x6ff - (0x6ff0)))] != ((void *0))' failed!

Tested with the following 2.6 images: 
older 2.6.10-1-386 (subversion 2), 
current 2.6.10-1-386 (subversion 10)
whatever the debian installer tries to put on the system when booted 
with 2.6 - most likely 
current 2.6.10-1-386 
2.6.9 (686 config altered to optimize for 386).

with the default 2.4.18 image from woody will boot normally

with 2.6.10-2 and 2.6.9 will boot normally if the libc6 is 2.3.2.ds1-20 or 
earlier.

I am looking at the changelog for ds1-21 and so far I have no idea what could 
have caused it.

-- System Information:
Debian Release: 3.1
  APT prefers testing
  APT policy: (500, 'testing')
Architecture: i386 (i686)
Kernel: Linux 2.4.18-bf2.4
Locale: LANG=C, LC_CTYPE=C (charmap=ANSI_X3.4-1968)

Versions of packages libc6 depends on:
ii  libdb1-compat 2.1.3-7The Berkeley database routines [gl

-- no debconf information


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#308792: After the last update Via C3 systems give assertion failed in ld.so at boot

2005-05-12 Thread Daniel Jacobowitz
On Thu, May 12, 2005 at 11:39:23AM +, Anton Ivanov wrote:
 Package: libc6
 Version: 2.3.2.ds1-21
 Severity: critical
 
 
 
 After the last update C3 Version 1
 
 with a kernel 2.6 image will fail on boot with:
 
 Inconsistency detected by ld.so: do_rel.h: 109 elf_dynamic_do_rel: Assertion 
 '(map-l_info[(34+0+(0x6ff - (0x6ff0)))] != ((void *0))' failed!
 
 Tested with the following 2.6 images: 
   older 2.6.10-1-386 (subversion 2), 
   current 2.6.10-1-386 (subversion 10)
   whatever the debian installer tries to put on the system when booted 
 with 2.6 - most likely 
 current 2.6.10-1-386 
   2.6.9 (686 config altered to optimize for 386).
 
 with the default 2.4.18 image from woody will boot normally
 
 with 2.6.10-2 and 2.6.9 will boot normally if the libc6 is 2.3.2.ds1-20 or 
 earlier.
 
 I am looking at the changelog for ds1-21 and so far I have no idea what could 
 have caused it.

That's:

#ifdef RTLD_BOOTSTRAP
  /* The dynamic linker always uses versioning.  */
  assert (map-l_info[VERSYMIDX (DT_VERSYM)] != NULL);
#else

The problem is not going to be anywhere near there.  That is a check on
the ld.so binary, which works elsewhere.  Probably your mmap is busted.
I do not see anything that would cause this in -21 either.


-- 
Daniel Jacobowitz
CodeSourcery, LLC


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#308792: After the last update Via C3 systems give assertion failed in ld.so at boot

2005-05-12 Thread Anton Ivanov
Daniel Jacobowitz wrote:
On Thu, May 12, 2005 at 11:39:23AM +, Anton Ivanov wrote:
 

Package: libc6
Version: 2.3.2.ds1-21
Severity: critical

After the last update C3 Version 1
with a kernel 2.6 image will fail on boot with:
Inconsistency detected by ld.so: do_rel.h: 109 elf_dynamic_do_rel: Assertion 
'(map-l_info[(34+0+(0x6ff - (0x6ff0)))] != ((void *0))' failed!

Tested with the following 2.6 images: 
	older 2.6.10-1-386 (subversion 2), 
	current 2.6.10-1-386 (subversion 10)
	whatever the debian installer tries to put on the system when booted with 2.6 - most likely 
current 2.6.10-1-386 
	2.6.9 (686 config altered to optimize for 386).

with the default 2.4.18 image from woody will boot normally
with 2.6.10-2 and 2.6.9 will boot normally if the libc6 is 2.3.2.ds1-20 or 
earlier.
I am looking at the changelog for ds1-21 and so far I have no idea what could have caused it.
   

That's:
#ifdef RTLD_BOOTSTRAP
 /* The dynamic linker always uses versioning.  */
 assert (map-l_info[VERSYMIDX (DT_VERSYM)] != NULL);
#else
The problem is not going to be anywhere near there.  That is a check on
the ld.so binary, which works elsewhere.  Probably your mmap is busted.
I do not see anything that would cause this in -21 either.
 

I tested with a few more versions and some alternative hardware.
2.4.27 also does not boot unless you turn off ACPI and APIC. If you turn 
them off it boots.

All 2.6 images and 2.4.27 boot OK on C3 V2.
It starts looking like an an interaction of specific hardware and 
drivers - C3 V1, CMD649 ide and a few others. I still have no idea why 
does it bomb out in mmap/ldso. With a hardware problem I would have 
expected it to barf much earlier and in a more consistent manner.

A.
--
La Châtelier's Law: 

 If some stress is brought to bear on a system in equilibrium, 
the equilibrium is displaced in the direction which tends to undo the 
effect of the stress.