Revisiting 2.4.X __alloc_pages: X-order allocation failed

From: Matthias Andree (matthias.andree@stud.uni-dortmund.de)
Date: Tue Feb 26 2002 - 20:55:51 EST

Hi,

I recently had troubles with excessively low I/O throughput when writing
a CD from an ISO image. The image file was on a reiserfs partition on an
IBM DCAS-32160 (USCSI), attached to an AHA2940Udontaskme (AIC7881U), the
writer is ATAPI 16x. Pentium-II/400, 256 MB, PIIX4 IDE chip (BX chip
set I believe). SuSE official 2.4.16 kernel as offered as update for
SuSE 7.3 -- I didn't check their patches, but their recent kernels were
based on -aa stuff AFAICS.

The low throughput showed in a big (16M) but empty (0 - 4%) cdrecord
FIFO, with the drive pausing during the write (didn't harm, burnproof
was on). 2400 kByte/s (for 16x CD writing) is far below the sustained
transfer rates which range from 4.7 at ID to 7.7 MB/s at OD.

This problem didn't show with vanilla 2.4.14.

I found a lot of __alloc_pages: X-order allocation failed in my logs,
with X from 1 to 3. Do these allocation failures impair I/O throughput
when reading stuff from a disk, tossing things through a
statically-allocated FIFO to a CD writer? If so, I might need to give
that box some other kernel. 

-- 
Matthias Andree

GPG encrypted mail welcome, unless it's unsolicited commercial email.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Thu Feb 28 2002 - 21:00:35 EST