Re: update/bdflush stall's pine/disk_IO while flushing ?

Mike Galbraith (mikeg@weiden.de)
Mon, 21 Dec 1998 06:16:47 +0100 (CET)


On Sun, 20 Dec 1998, Arjan Filius wrote:

> Hello,
>
> I noticed to many times my pine will block while reading Email (such as
> kernel-digest, which becommes huge), and bdflush gets active. With vmstat
> i see then a huge raise in io/bo, and when i try to scroll trough the
> Email message, pine just waits and waits, about 10 or more seconds.

No help, but some additional data..

I see the same thing. I just did a trace immediately after such an event
and most of the time the kernel is in do_check_pgt_cache.

Here, I'm (p955 is pine) doing next/scroll through it until the 'storm'
hits. System is otherwise idle.

%TIME TOTAL USECS AVG/CALL CALLS ADDR
0.4811% 13909.49 1.62 8573 c012a15b refile_buffer p955
0.4928% 14245.78 1.66 8569 c01295e7 find_buffer p955
0.6815% 19700.86 9850.43 2 c0106e76 hard_idle p0
0.7102% 20530.49 5.64 3643 c01a949b make_request p955
0.8344% 24123.40 47.12 512 c01088f6 ret_from_sys_call p955
2.5337% 73250.23 581.35 126 c0107697 __switch_to p0
10.3459% 299098.64 946.51 316 c01122db schedule p0
78.8774% 2280343.30 6000.90 380 c010f823 do_check_pgt_cache p0
121909 entries; 2890998.17 total usecs.

Since I'm having no problems with anything except pine, I've been assuming
that pine is doing something wrong and just ignoring it.

-Mike

-
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/