* Benjamin Herrenschmidt [off-list ref] wrote:

On Tue, 2009-09-22 at 09:48 +1000, Paul Mackerras wrote:

quoted

This fixes two places in the powerpc perf_event (perf_counter) code 
where 'list_entry' needs to be changed to 'group_entry', but were 
missed in commit 65abc865 ("perf_counter: Rename list_entry -> 
group_entry, counter_list -> group_list").

Oops, indeed - queued up the fix and will send it to Linus shortly - 
thanks!

Ingo: This is becoming a recurring one now... powerpc build upstream 
is broken approx everyday by some new perfctr build breakage.

You really aren't build testing other architectures than x86 right ?

On the contrary - i am build testing every architecture on a daily 
basis. (and sometimes i do it multiple times a day - yesterday i did 5 
cross builds during the rename) In fact i am testing more architectures 
than linux-next does.

Here's the log of the test i ran yesterday before i sent those bits to 
Linus:

testing 24 architectures.
                                 (warns)               (warns)
testing      alpha:  -git:  pass (   24),  -tip:  pass (   24)
testing        arm:  -git:  fail (   11),  -tip:  fail (   13)
testing   blackfin:  -git:  pass (    3),  -tip:  pass (    3)
testing       cris:  -git:  fail (   34),  -tip:  pass (   20)
testing        frv:  -git:  fail (   13),  -tip:  fail (   13)
testing      h8300:  -git:  fail (  441),  -tip:  fail (  185)
testing       i386:  -git:  pass (    2),  -tip:  pass (    5)
testing       ia64:  -git:  fail (  172),  -tip:  pass (  160)
testing       m32r:  -git:  pass (   39),  -tip:  pass (   39)
testing       m68k:  -git:  pass (   42),  -tip:  pass (   42)
testing  m68knommu:  -git:  fail (   80),  -tip:  fail (   80)
testing microblaze:  -git:  fail (   14),  -tip:  fail (   14)
testing       mips:  -git:  pass (    6),  -tip:  pass (    6)
testing    mn10300:  -git:  fail (   10),  -tip:  fail (   10)
testing     parisc:  -git:  pass (   26),  -tip:  pass (   26)
testing    powerpc:  -git:  fail (   36),  -tip:  fail (   45)
testing       s390:  -git:  pass (    6),  -tip:  pass (    6)
testing      score:  -git:  fail (   13),  -tip:  fail (   13)
testing         sh:  -git:  fail (   22),  -tip:  fail (   19)
testing      sparc:  -git:  pass (    3),  -tip:  pass (    3)
testing         um:  -git:  pass (    3),  -tip:  pass (    3)
testing     xtensa:  -git:  fail (   46),  -tip:  fail (   46)
testing     x86-64:  -git:  pass (    0),  -tip:  pass (    0)
testing     x86-32:  -git:  pass (    0),  -tip:  pass (    0)

In fact there are architectures that dont build in Linus's tree and 
build in -tip:

testing       cris:  -git:  fail (   34),  -tip:  pass (   20)

Because not only do i test every architecture i also try to fix upstream 
bugs on non-x86 pro-actively. See for example this upstream fix:

 8d7ac69: Blackfin: Fix link errors with binutils 2.19 and GCC 4.3

Nevertheless you are right that i should have caught this particular 
PowerPC build bug - i missed it - sorry about that!

Thanks,

	Ingo