Re: AUFS

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Tue, 11 Dec 2001 16:18:48 +0100

I dont mind if it fixes your problems. We have accounting for the number of
pending opens also.

However, I would still like to understand what went wrong in the other
approach. I am afraid that it might be a deeper problem.. (possibly with my
brain)

Regards
Henrik

On Tuesday 11 December 2001 10.09, Adrian Chadd wrote:
> This seems to fix the FD accounting.
> Don't know why .. but it just makes sense to do it this way.
>
> Do you mind if I commit this?
>
>
>
> Adrian
>
> Index: src/fs/aufs/store_io_aufs.c
> ===================================================================
> RCS file: /server/cvs-server/squid/squid/src/fs/aufs/store_io_aufs.c,v
> retrieving revision 1.14
> diff -u -r1.14 store_io_aufs.c
> --- src/fs/aufs/store_io_aufs.c 2001/10/24 07:45:37 1.14
> +++ src/fs/aufs/store_io_aufs.c 2001/12/11 03:28:07
> @@ -71,7 +71,6 @@
> #else
> storeAufsOpenDone(fd, sio, fd, 0);
> #endif
> - store_open_disk_fd++;
> return sio;
> }
>
> @@ -126,7 +125,6 @@
> #else
> storeAufsOpenDone(fd, sio, fd, 0);
> #endif
> - store_open_disk_fd++;
>
> /* now insert into the replacement policy */
> storeAufsDirReplAdd(SD, e);
> @@ -284,6 +282,7 @@
> storeAufsIOCallback(sio, DISK_ERROR);
> return;
> }
> + store_open_disk_fd++;
> aiostate->fd = fd;
> commSetCloseOnExec(fd);
> fd_open(fd, FD_FILE, storeAufsDirFullPath(INDEXSD(sio->swap_dirn),
> sio->swap_filen, NULL)); @@ -409,12 +408,12 @@
> cbdataUnlock(their_data);
> aiostate->fd = -1;
> cbdataFree(sio);
> - store_open_disk_fd--;
> if (fd < 0)
> return;
> debug(78, 3) ("%s:%d\n", __FILE__, __LINE__);
> aioClose(fd);
> fd_close(fd);
> + store_open_disk_fd--;
> debug(78, 3) ("%s:%d\n", __FILE__, __LINE__);
> }
Received on Tue Dec 11 2001 - 08:18:33 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:40 MST