Skip to contentSkip to conversation

v3.4-rc2 out-of-memory problems (was Re: 3.4-rc1 sticks-and-crashs)

DISCUSSION werner 2 participants 3 messages Apr 10, 2012 Stable branch

lore
Share & tools
lore

werner

author
#0 lore raw

At least until now, and also tested hard by starting and stopping several memory-consuming operations, I'm happy to can inform that the computer didnt yet crash again, and that also slownessnesses what I observed under 3.3 (but without crashs) don't occure and kmemleak (or a similar name) what sometimes started and I had to shut down under 3.3 for recover normal speed, I dont see longer, so that the the patch suggested by D.R. and L.T. cured the crashs (at least phenomenologically). For be sure, however, it's still too early, only if the computer dont crash until tomorrow.

If there is found an agreement about something else what's senseful to test out, I can do that.

wl
---
Professional hosting for everyone - http://www.host.ru

#1 lore raw
Quoted 4 lines · werner

At least until now, and also tested hard by starting and stopping several memory-consuming operations, I'm happy to can inform that the computer didnt yet crash again, and that also slownessnesses what I observed under 3.3 (but without crashs) don't occur

That could be due to a few VM patches which I wrote, that went in through -mm.

I am very interested in whether people do find a way to break the VM with those patches, in ways that used to work before.

If you find any, please let me know so I can fix them before the 3.4 kernel comes out.

If everything you try works better than before, I'm not going to complain about good news :)

Signature · 1 line
All rights reversed
author
#2 lore raw

Rik van Riel :
To avoid misunderstandings, that improvement is WITH / INCLUDING the patch suggested by D.R and L.T., namely: put + free_task(task); after line 78 in subroutine lowmemorykiller.c . At least phenomenologically, it looks that it resolved the problem
I dont understanding so much about programming and about the problem (and "android" is neither running on my own computer nor on my coffee-machine, but I enable everything just because I don't know what hardware have the people which use my distro and compiled kernels) so that I don't understand all details of the discusion, but if people come to an agreement what I could test further, one can mail it me and explain it me slowly, then I test it out. wl

==========

Quoted 13 lines · On Mon, 09 Apr 2012 21:51:58 -0400 Rik van Riel <[email protected]> wrote:
Quoted 4 lines · werner

At least until now, and also tested hard by starting and stopping several memory-consuming operations, I'm happy to can inform that the computer didnt yet crash again, and that also slownessnesses what I observed under 3.3 (but without crashs) don't occur

That could be due to a few VM patches which I wrote, that went in through -mm.

I am very interested in whether people do find a way to break the VM with those patches, in ways that used to work before.

Quoted 10 lines

If you find any, please let me know so I can fix them before the 3.4 kernel comes out.

If everything you try works better than before, I'm not going to complain about good news :)

Signature · 1 line
All rights reversed

"werner" <[email protected]>
---
Professional hosting for everyone - http://www.host.ru