Skip to contentSkip to conversation

PATCH: Precise Accounting for 2.6.7

DISCUSSION Andi Kleen 2 participants 2 messages Jun 23, 2004 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Andi Kleen

author
#1 lore raw
Quoted 4 lines · Timm Morten Steinbeck

Hi,

we have ported our x86 precise accounting patch from the 2.4 kernel series to 2.6.7.

[...]

On many SMP systems it will not be very precise after longer uptimes because the TSCs of the CPUs drift away noticeable and timestamps from different CPUs cannot be really compared.

-Andi

#2 lore raw

Am Mit, 2004-06-23 um 16.54 schrieb Andi Kleen:

Quoted 4 lines
Quoted 2 lines · Timm Morten Steinbeck

Hi,

Quoted 8 lines
Quoted 2 lines

we have ported our x86 precise accounting patch from the 2.4 kernel series to 2.6.7.

[...]

On many SMP systems it will not be very precise after longer uptimes because the TSCs of the CPUs drift away noticeable and timestamps from different CPUs cannot be really compared.

As far as we understand the problem you describe we do not see this as a problem.
The patch keeps track of the TSCs of each CPU separately and only uses the differences of values read on the same CPU. At least that was our intention, but maybe we overlooked something. Can you point us to the places in our patch where this problem occurs?

Best regards
Arne & Timm

********************************************************************
Timm M. Steinbeck & Kirchhoff Institute for Physics
Arne Wiebalck Computer Science/Computer Engineering
e-mail: INF 227, Room 3.315
timm.steinbeck AT kip.uni-heidelberg.de D-69120 Heidelberg
arne.wiebalck AT kip.uni-heidelberg.de
web: Tel.: (+49) 6221/54-9816
http://www.kip.uni-heidelberg.de Fax.: (+49) 6221/54-9809
********************************************************************