i recently got a gig of ram for cheap and installed it and recompiled my gentoo-dev-sources kernel (2.6.11-gentoo-r6). with HIGHMEM enabled the machine needs about 15 minutes to come up (vs <1min w/o HIGHMEM). when i recompile the kernel without HIGHMEM or boot the kernel with mem=<884MB everything's as usual. the same problem appears when using eg ck-sources, mm-sources etc. ... (only tried 2.6-series kernels) the slow part actualy starts with INIT: ... Reproducible: Always Steps to Reproduce: 1.choose HIGHMEM option in kernel 2.compile and install kernel 3.reboot Actual Results: every applications has a 5-10x times longer start up time. general cpu performance is ok (even better than w/ 1GB lowmem patch!). hdparm -tT reports 1.9GB/s with HIGHMEM and 1.5GB/s with lowmem patch, disc read performance with HIGHMEM is ~ 68.5MB/s vs 58MB/s with 1GB-lowmem vs 68.5MB/s w/o lowmem(default) and w/o HIGHMEM. *It seems that a lot of ppl have this kind of error - see thread @ forums: http://forums.gentoo.org/viewtopic-p-2443306.html#2443306 *search on google for `HIGHMEM slow boot'
am i right here with posting kernel bugs - or should i stick with real kernel developers over at kernel.org?
Here is ok, see http://dev.gentoo.org/~dsd/gentoo-sources/bugs.htm Please try and reproduce with 2.6.12-rc5
(In reply to comment #2) > Here is ok, see http://dev.gentoo.org/~dsd/gentoo-sources/bugs.htm > Please try and reproduce with 2.6.12-rc5 tried it with 2.6.12-rc4 - nothing changed. i'll try rc5 asap.
(In reply to comment #3) > (In reply to comment #2) > > Here is ok, see http://dev.gentoo.org/~dsd/gentoo-sources/bugs.htm > > Please try and reproduce with 2.6.12-rc5 > > tried it with 2.6.12-rc4 - nothing changed. > i'll try rc5 asap. what's realy intresing is that putc/getc and disc I/O performance (!hdparm) is realy bad: highmem > 1024MB Seeker 1...Seeker 2...Seeker 3...start 'em...done...done...done... -------Sequential Output-------- ---Sequential Input-- --Random-- -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks--- Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU 100 843 99.9 214295 78.5 766914 100.3 731 99.9 2364623 101.6 97878.5 185.9 highmem, booted with 880MB Seeker 1...Seeker 2...Seeker 3...start 'em...done...done...done... -------Sequential Output-------- ---Sequential Input-- --Random-- -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks--- Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU 100 45898 99.7 393189 100.2 894995 94.4 50903 100.0 2453987 100.6 147552.5 184.4 putc's are 50 times slower!
In that case, it looks like an upstream bug. Please file a bug report at http://bugzilla.kernel.org and post the resultant URL here.