http://defect.opensolaris.org/bz/show_bug.cgi?id=6334
Eric Saxe <eric.saxe at sun.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Summary|PG walker reports UNKNOWN |Group walker gives up too
|for all PGs on nehalem |easily with sparse groups
Status|ACCEPTED |FIXINPROGRESS
--- Comment #2 from Eric Saxe <eric.saxe at sun.com> 2009-02-06 10:38:30 ---
I just BFUed the tests system in question, and the output is now fine. This was
likely due to a mismatch between the mdb module and the kernel with respect to
their understanding of "struct pghw"...
> *pg_hw::walk group|::walk group|::pg
PGID ADDR PHYSID CLASS HARDWARE #CPUs
0 ffffff02520120a0 0 cmt ipipe 2
4 ffffff025456b338 1 cmt ipipe 2
6 ffffff02547e8dc0 2 cmt ipipe 2
8 ffffff02547e8780 3 cmt ipipe 2
10 ffffff0254a41f08 8 cmt ipipe 2
14 ffffff0254a416e8 9 cmt ipipe 2
16 ffffff0254a410a8 10 cmt ipipe 2
18 ffffff0254b9c970 11 cmt ipipe 2
1 ffffff0252012000 0 cmt cache 8
11 ffffff0254a41e68 1 cmt cache 8
2 ffffff0253e06f08 0 cmt chip 8
12 ffffff0254a41dc8 1 cmt chip 8
3 ffffff0253e06e68 0 cmt idle_pwr 2
5 ffffff025456b298 1 cmt idle_pwr 2
7 ffffff02547e8d20 2 cmt idle_pwr 2
9 ffffff02547e86e0 3 cmt idle_pwr 2
13 ffffff0254a41d28 8 cmt idle_pwr 2
15 ffffff0254a41648 9 cmt idle_pwr 2
17 ffffff0254a41008 10 cmt idle_pwr 2
19 ffffff0254b9c8d0 11 cmt idle_pwr 2
However, there is another issue here. The group walker assumes that the
elements in a group_t are packed, which isn't the case if groups are added at
particular indices (so some may be NULL). The walker terminates in this case,
which is why despite the output above being coherent, it's not complete (the
active power domain groupings are missing. So rather than closing out this bug,
and filing another, I'm going to morph this into the new bug to fix the group
walker.
--
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
