Skip to contentSkip to conversation

Announce: kdb v4.4 is available for kernel 2.6.10

DISCUSSION Keith Owens 2 participants 2 messages Dec 25, 2004 – Dec 27, 2004

lore
Share & tools
lore
#0 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

KDB (Linux Kernel Debugger) has been updated.

ftp://oss.sgi.com/projects/kdb/download/v4.4/ ftp://ftp.ocs.com.au/pub/mirrors/oss.sgi.com/projects/kdb/download/v4.4/

Current versions are :-

kdb-v4.4-2.6.10-common-1.bz2
kdb-v4.4-2.6.10-i386-1.bz2
kdb-v4.4-2.6.10-ia64-1.bz2 kdb-v4.4-2.6.9-rc2-x86-64-1.bz2 (may or may not work with 2.6.10).

Changelog extract since kdb-v4.4-2.6.9-common-1.

2004-12-25 Keith Owens <[email protected]>

* Add kobject command.
* Ignore low addresses and large offsets in kdbnearsym().
* Console updates for sn2 simulator.
* kdb v4.4-2.6.10-common-1.

2004-12-07 Keith Owens <[email protected]>

* kdb v4.4-2.6.10-rc3-common-1.

2004-11-23 Keith Owens <[email protected]>

* Remove warning message from kdb_get_one_user_page(), it was too noisy.
* kdb v4.4-2.6.10-rc2-common-1.

2004-11-02 Keith Owens <[email protected]>

* Build with kdb patch applied but CONFIG_KDB=n.
* kdb v4.4-2.6.10-rc1-common-2.

2004-10-29 Keith Owens <[email protected]>

* Handle new compression scheme for kallsyms.
* Handle move of DEAD and ZOMBIE for task->state to task->exit_state.
* Tweak the concept of a valid kernel address to get all symbols, including the symbols in the ia64 gate page.
* kdb v4.4-2.6.10-rc1-common-1.

2004-10-21 Keith Owens <[email protected]>

* Handle variable size for the kernel log buffer.
* kdb v4.4-2.6.9-common-2.

Changelog extract since kdb-v4.4-2.6.9-i386-1.

2004-12-25 Keith Owens <[email protected]>

* kdb v4.4-2.6.10-i386-1.

2004-12-07 Keith Owens <[email protected]>

* kdb v4.4-2.6.10-rc3-i386-1.

2004-11-23 Keith Owens <[email protected]>

* Coexist with asmlinkage/fastcall changes.
* kdb v4.4-2.6.10-rc2-i386-1.

2004-10-29 Keith Owens <[email protected]>

* Handle change defintions for hard and soft irq context.
* Make stack switch in kdb backtrace look more like the oops output.
* kdb v4.4-2.6.10-rc1-i386-1.

Changelog extract since kdb v4.4-2.6.9-ia64-1.

2004-12-25 Keith Owens <[email protected]>

* Add cpuinfo command.
* kdb-v4.4-2.6.10-ia64-1.

2004-12-07 Keith Owens <[email protected]>

* Clean up error path in kdba_mca_init.
* kdb-v4.4-2.6.10-rc3-ia64-1.

2004-11-15 Keith Owens <[email protected]>

* kdb-v4.4-2.6.10-rc2-ia64-1.

2004-10-29 Keith Owens <[email protected]>

* kdb-v4.4-2.6.10-rc1-ia64-1.

ps. Bah, Hum{de}bug!

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Exmh version 2.1.1 10/15/1999

iD8DBQFBzVpji4UHNye0ZOoRAmy+AJ4iSaztTqGLjr+Ck0X8+TMdXB41IQCghc/P p7GtfCEOmVPDj/SVHdecFyw=oO+Z
-----END PGP SIGNATURE-----

#1 lore raw
Quoted 4 lines · Keith Owens

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

KDB (Linux Kernel Debugger) has been updated.

Hi,

I browsed the patch quickly to check for kallsyms uses, and validate them, and it generally seems correct.

There is however one comment that makes me wonder:
*2.6 kallsyms has a "feature" where it unpacks the name into a string. *If that string is reused before the caller expects it then the caller *sees its string change without warning.

kallsyms_lookup always uses the buffer passed to it in the case the symbol is a kernel symbol, as opposed to a module symbol, and so it is not responsible for the buffer.

So this probably only happens when a module symbol is returned directly from its symbol table, and then the module is unloaded (or something like that).

Later there is another comment:
* Another 2.6 kallsyms "feature". Sometimes the sym_name is
* set but the buffer passed into kallsyms_lookup is not used,
* so it contains garbage.

It seems to be the same problem. If we modify kallsyms_lookup to always use the buffer passed, even if the symbol comes from a module, maybe we could solve both problems with just one change.

On the downside, a caller that just wants to print the name, would pay an unnecessary string copy.

On the upside, this would make the interface more coherent with standard C functions like strcpy, where the buffer passed is always the buffer returned.

So, is it worth the change?

Signature · 4 lines
Paulo Marques - www.grupopie.com

"A journey of a thousand miles begins with a single step."
Lao-tzu, The Way of Lao-tzu