Date: Fri, 24 Sep 2004 12:12:56 -0400 From: David Schultz <das@FreeBSD.ORG> To: Pawel Jakub Dawidek <pjd@FreeBSD.ORG> Cc: Current <freebsd-current@FreeBSD.ORG> Subject: Re: panic: swap_pager_isswapped: failed to locate all swap meta blocks Message-ID: <20040924161255.GA699@VARK.MIT.EDU> In-Reply-To: <20040924122508.GG9550@darkness.comp.waw.pl> References: <7m7jqjhojv.wl@black.imgsrc.co.jp> <20040924122508.GG9550@darkness.comp.waw.pl>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Sep 24, 2004, Pawel Jakub Dawidek wrote:
> On Fri, Sep 24, 2004 at 08:52:20PM +0900, Jun Kuriyama wrote:
> +>
> +> This is current as of 2004.09.23.23.00.00+00 (when I hit "shutdown -r
> +> now").
> +>
> +> -----
> +> Shutting down local daemons:.
> +> Writing entropy file:.
> +> panic: swap_pager_isswapped: failed to locate all swap meta blocks
> +> cpuid = 1
> +> KDB: enter: panic
> +> [thread 100138]
> +> Stopped at kdb_enter+0x2b: nop
> +> db> tr
> +> kdb_enter(c068a149) at kdb_enter+0x2b
> +> panic(c069dbcc,1000000,20000020,0,c43ee000) at panic+0x127
> +> swap_pager_isswapped(c42c44a4,c392e880,c42c44a4,0,c069dec7,18c) at swap_pager_isswapped+0xf6
> +> vm_proc_swapin_all(c392e880,c072d540,0,c069da42,8b6) at vm_proc_swapin_all+0x6a
> +> swapoff(c3b8bc00,ecda9d14,1,1,296) at swapoff+0x1b6
> +> syscall(2f,2f,2f,8049368,1) at syscall+0x213
> +> Xint0x80_syscall() at Xint0x80_syscall+0x1f
> +> --- syscall (424, FreeBSD ELF32, swapoff), eip = 0x280c066b, esp = 0xbfbfee0c, ebp = 0xbfbfee28 ---
> +> -----
> +>
> +> I have a core if you want.
Sorry, I just committed a patch that I believe will fix this.
> This is indirectly caused by me. I added 'stop' method to rc.d/swap1 script
> and now swapoff(8) is done on every shutdown(8) run.
>
> Will it be possible to fix it or should I made this optional?
One thing worth pointing out is that the present swapoff(2)
implementation can be pretty slow if there's a lot of stuff
swapped out. It swaps data in synchronously, one page at a time.
It would not be difficult to improve on this, but it isn't a
priority. You might want to look at how fast it is and see if
this is acceptable.[1]
One way to speed it up would be to wait until after most userland
processes have been killed, since most of the data won't need to
be swapped back in in that case. In fact, this is an improvement
for another reason, namely, there might not be enough physical
memory for swapoff() to work otherwise.
[1] On the other hand, one advantage of having it invoked on shutdown
is that it will be tested routinely. ;-)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040924161255.GA699>