Messages in this thread Patch in this message | | | Date | Fri, 30 Apr 1999 15:19:35 +0200 (CEST) | From | Andrea Arcangeli <> | Subject | Re: 2.2.7_andrea2.bz2 |
| |
On Fri, 30 Apr 1999, Ben McCann wrote:
>Great Job, Andrea!
Thanks!! ;))
>Linus (if you're reading this), would you consider Andrea's VM tuning >changes for an up-coming point release of 2.2?
2.2.6_andrea2 was faster than 2.2.7_andrea2. So before start stabilizing the code I am reasearching the reasons for this performances-loss. I think this patch could be the one that will give us again 2.2.6_andrea2 performances. The patch is against 2.2.7_andrea2:
Index: fs/buffer.c =================================================================== RCS file: /var/cvs/linux/fs/buffer.c,v retrieving revision 1.1.2.88 diff -u -r1.1.2.88 buffer.c --- buffer.c 1999/04/29 18:54:03 1.1.2.88 +++ buffer.c 1999/04/30 02:24:32 @@ -764,7 +764,8 @@ */ if (too_many_dirty_buffers()) { - flush_dirty_buffers(); + if (nr_free_pages <= freepages.high) + flush_dirty_buffers(); if (too_many_dirty_buffers()) wakeup_bdflush(); }
I am not completly sure though. I removed the code above after 2.2.6_andrea2 in order to have a saner VM but it wasn't a bug at all. So now I reinserted it and I am waiting a feedback from Harvey who is doing a great work (as you all) in testing my code.
The good thing is that 2.2.7_andrea2 is rock solid while previous kernels had bugs (swapoff -a was going to corrupt memory... excuse me but I missed a bit of code in unuse_pte, if just fixed it in 2.2.7_andrea2 though).
If there will be interest in merging my code in 2.2.x I'll be _glad_ ;) to provide patches for only the relevant/major VM changes for inclusion.
Andrea Arcangeli
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |