Messages in this thread | | | Date | Mon, 23 Feb 1998 20:54:22 -0800 (PST) | From | Trevor Johnson <> | Subject | Re: [patch] large swap areas |
| |
I tried your patches, and had only a minor problem, which seems related to it. I booted with mem=6M, then ran X, Mosaic, and several other programs to get about 45 MB into the 1.2 GB swap file I made. Because I tried to switch between X and text, the video display became corrupted, so I rebooted (using Ctrl-Alt-Del). I tried booting with mem=5M, but my system was constantly thrashing even though it didn't have any swap enabled. I wasn't able to log in, or to reboot with Ctrl-Alt-Del. Doing a hard reboot and trying mem=6M again, I logged in as root and did the following:
Script started on Mon Feb 23 18:24:56 1998 # cat /proc/meminfo total: used: free: shared: buffers: cached: Mem: 4878336 4554752 323584 233472 135168 217088 Swap: 0 0 0 MemTotal: 4764 kB MemFree: 316 kB MemShared: 228 kB Buffers: 132 kB Cached: 212 kB SwapTotal: 0 kB SwapFree: 0 kB # /usr/src/mkswap /dev/hdd /usr/src/mkswap: warning: large swap area (10 signature pages) Setting up swapspace, size = 1282494464 bytes # swapon /dev/hdd swapon: /dev/hdd: Cannot allocate memory # cat /proc/meminfo total: used: free: shared: buffers: cached: Mem: 4878336 4567040 311296 225280 131072 225280 Swap: 0 0 0 MemTotal: 4764 kB MemFree: 304 kB MemShared: 220 kB Buffers: 128 kB Cached: 220 kB SwapTotal: 0 kB SwapFree: 0 kB # exit exit
Script done on Mon Feb 23 18:26:02 1998 [I left "script" to free up enough RAM to run "swapon".] # swapon /dev/hdd Adding Swap: 1252400k swap-space (priority -4) # cat /proc/meminfo total: used: free: shared: buffers: cached: Mem: 4878336 4468736 409600 811008 131072 724992 Swap: 1282457600 471040 1281986560 MemTotal: 4764 kB MemFree: 400 kB MemShared: 792 kB Buffers: 128 kB Cached: 708 kB SwapTotal: 1252400 kB SwapFree: 1251940 kB # modprobe aha1542 Configuring Adaptec (SCSI-ID 7) at IO:334, IRQ 10, DMA priority 6 scsi1 : Adaptec 1542 scsi : 1 host. Vendor: ARCHIVE Model: Python 28388-XXX Rev: 5.45 Type: Sequential-Access ANSI SCSI revision: 02 # modprobe st # /usr/src/mkswap /dev/st0 /usr/src/mkswap: warning: large swap area (9 signature pages) Setting up swapspace, size = 1073737728 bytes [The patched mkswap decided a 90 m DDS tape could hold 1 GB - 4 kB. Its actual capacity is about (exactly?) 2 GB.] # swapon /dev/st0 swapon: /dev/st0: Invalid argument [Hey, I had to try. Then I logged in as myself and started some programs.] % ps uaxw USER PID %CPU %MEM SIZE RSS TTY STAT START TIME COMMAND root 1 0.0 0.0 832 0 ? SW 18:20 0:03 (init) root 2 0.0 0.0 0 0 ? SW 18:20 0:00 (kflushd) root 3 9.0 0.0 0 0 ? SW< 18:20 12:36 (kswapd) root 8 0.0 0.0 8 0 ? SW 18:20 0:00 (update) root 23 0.0 0.0 796 0 ? SW 18:22 0:00 (kerneld) root 57 0.0 0.0 1020 0 ? SW 18:22 0:01 (syslogd) root 60 0.0 0.0 1048 0 ? SW 18:22 0:00 (klogd) root 61 0.0 0.0 1384 0 ? SW 18:22 0:01 (named) root 64 0.0 0.0 916 0 ? SW 18:22 0:00 (inetd) root 70 0.0 0.0 936 0 ? SW 18:22 0:01 (cron) root 74 0.0 0.0 1408 0 ? SW 18:22 0:00 (sendmail) root 77 0.0 0.0 804 0 ? SW 18:22 0:01 (gpm) root 78 0.1 0.0 1956 0 1 SW 18:23 0:09 (tcsh) root 87 0.0 0.0 780 0 10 SW 18:23 0:00 (mingetty) root 127 0.0 0.0 0 0 1 SW 18:30 0:00 (scsi_eh_1) root 145 0.1 1.3 1208 64 S1 S 18:32 0:14 pppd root 395 0.0 0.0 780 0 7 SW 20:37 0:00 (mingetty) root 396 0.0 0.0 780 0 9 SW 20:37 0:00 (mingetty) trevor 79 0.0 0.0 1912 0 2 SW 18:23 0:00 (tcsh) trevor 81 0.0 0.0 1916 0 4 SW 18:23 0:00 (tcsh) trevor 82 0.0 0.0 2060 0 5 SW 18:23 0:00 (tcsh) trevor 83 0.0 0.0 2060 0 6 SW 18:23 0:00 (tcsh) trevor 85 0.0 0.0 2056 0 8 SW 18:23 0:04 (tcsh) trevor 88 0.0 0.0 1920 0 11 SW 18:23 0:01 (tcsh) trevor 113 0.5 0.0 15584 0 2 SW 18:26 0:41 (pine) trevor 292 0.0 0.0 1920 0 3 SW 19:34 0:02 (tcsh) trevor 315 0.8 0.0 114252 0 4 SW 19:45 0:28 (emacs) trevor 323 1.0 0.0 114252 0 5 SW 19:50 0:29 (emacs) trevor 331 1.1 0.0 114252 0 6 SW 19:55 0:30 (emacs) trevor 342 10.7 0.0 891600 0 11 .W 20:02 3:58 (memgrabber) trevor 399 17.0 8.3 972 400 3 R 20:38 0:00 ps uaxw [The "memgrabber" is a small program by Ian Main.] % cat /proc/meminfo total: used: free: shared: buffers: cached: Mem: 4878336 4665344 212992 401408 126976 503808 Swap: 1282457600 1267159040 15298560 MemTotal: 4764 kB MemFree: 208 kB MemShared: 392 kB Buffers: 124 kB Cached: 492 kB SwapTotal: 1252400 kB SwapFree: 14940 kB % cat /proc/uptime 9224.32 7754.48
About 1 GB into swap, I got an error while working in PINE:
sh: fork: Cannot allocate memory
This error may have been the result of the resource limits I have set:
% limit cputime unlimited filesize 1048576 kbytes datasize unlimited stacksize 8192 kbytes coredumpsize 0 kbytes memoryuse unlimited descriptors 1024 memorylocked unlimited maxproc 256 openfiles 1024
However, the fact that it came near the 1 GB mark seems a bit suspicious. Here's what /proc/meminfo looked at shortly after the error:
total: used: free: shared: buffers: cached: Mem: 4878336 4595712 282624 643072 126976 540672 Swap: 1282457600 1061142528 221315072 MemTotal: 4764 kB MemFree: 276 kB MemShared: 628 kB Buffers: 124 kB Cached: 528 kB SwapTotal: 1252400 kB SwapFree: 216128 kB
As a side note, why are the redundant tables back in /proc/meminfo? Wasn't it simplified around 2.1.41? ___ Trevor Johnson
| |