Bug 317937 – dev-lang/python-{2.6.5-r2,3.1.2-r3} fail to build with Linux <2.6.9 headers

• Status: RESOLVED FIXED
• Product: Gentoo/Alt

[Description Jeremy Olexa (darkside) (RETIRED) 2010-04-30 20:22:19 UTC]

Platform is sles 9 ia64:
Linux hostname 2.6.5-7.191-default #1 SMP Tue Jun 28 14:58:56 UTC 2005 ia64 ia64 ia64 GNU/Linux

Python needs a patch or else it will try to use an undefined constant. It appears PR_SET_NAME is defined around the 2.6.9 kernel era.

[Comment 1 Jeremy Olexa (darkside) (RETIRED) 2010-04-30 20:23:11 UTC]

Created attachment 229841 [details]
python-2.6.5-python.c-PR_SET_NAME.patch

workaround that I used to successfully compile.

[Comment 2 Fabian Groffen 2010-05-02 11:39:56 UTC]

61_all_process_data.patch:+             prctl(PR_SET_NAME, process_name);

so the problem is introduced by a Gentoo patch.

[Comment 3 Fabian Groffen 2010-05-02 11:41:24 UTC]

61 is the GENTOO_PYTHON_* stuff patch.

@python: what you think of Jeremy's patch?

[Comment 4 Jeremy Olexa (darkside) (RETIRED) 2010-05-12 22:13:27 UTC]

Hi, I know this is low priority but I can't submit upstream for a revert to a Gentoo patch. Guidance?

[Comment 5 John Ross Hunt 2010-06-01 22:26:29 UTC]

I'm getting the same error on on ppc embedded: Linux dsmg600 2.4.21-pre4 #11 Sat Oct 3 11:50:34 CDT 2009 ppc 82xx Sandpoint GNU/Linux

Upgrading the kernel isn't much of an option unless I solder a serial port to the board and learn how to flash u-boot.

This appears to be a function that sets the process name, no?  Is it really critical for < 2.6.9 kernels?  Perhaps some of the newer prctl constants can be #ifdef'ed out for older kernels.  My concern is that the PR_SET_NAME call will have unpredictable results for older systems that don't define this function.

The prctl.h header is very sparse on my system (I'll include it if it helps).

[Comment 6 Fabian Groffen 2010-06-03 18:25:35 UTC]

*** Bug 322653 has been marked as a duplicate of this bug. ***

[Comment 7 Fabian Groffen 2010-06-03 18:28:29 UTC]

Can't we just axe this patch, or apply the one from Jeremy in Prefix?  Knowing the python team policies, the patch probably went upstream, and got accepted, so just another case of where we have to fight python upstream's annoyingness.

[Comment 8 Arfrever Frehtes Taifersar Arahesis (RETIRED) 2010-06-03 18:47:28 UTC]

This patch is probably wrong. We will apply better patch.

[Comment 9 Arfrever Frehtes Taifersar Arahesis (RETIRED) 2010-06-04 15:04:52 UTC]

(In reply to comment #8)

Apparently Jeremy Olexa mistested a test file...

[Comment 10 Arfrever Frehtes Taifersar Arahesis (RETIRED) 2010-06-29 03:51:27 UTC]

2.7, 3.2: r345

[Comment 11 Arfrever Frehtes Taifersar Arahesis (RETIRED) 2010-06-29 05:25:12 UTC]

Fixed in 2.7_pre20100627 and 3.2_pre20100627.

[Comment 12 Arfrever Frehtes Taifersar Arahesis (RETIRED) 2010-07-12 16:51:28 UTC]

2.6, 3.1: r356

[Comment 13 Arfrever Frehtes Taifersar Arahesis (RETIRED) 2010-07-12 19:32:44 UTC]

Fixed in 2.6.5-r3 and 3.1.2-r4.