Date: Tue, 03 Feb 1998 16:44:05 -0800 From: Kirk McKusick <mckusick@McKusick.COM> To: Julian Elischer <julian@whistle.com> Cc: current@FreeBSD.ORG Subject: Re: sb being written to r/o disks. Message-ID: <199802040044.QAA09193@flamingo.McKusick.COM> In-Reply-To: Your message of "Tue, 03 Feb 1998 16:30:41 PST." <34D7B6B0.31DFF4F5@whistle.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Sender: julian@whistle.com Date: Tue, 03 Feb 1998 16:30:41 -0800 From: Julian Elischer <julian@whistle.com> Organization: Whistle Communications To: mckusick@mckusick.com, current@freebsd.org Subject: sb being written to r/o disks. hmmmm it's amazing what new debugging info brings to light! I noticed this with the mount from the soft-updates integration. notice the two extra fields in 'mount'. # mount /dev/wd0e on / (local, read-only, writes: sync 4 async 0) /dev/wd0h on /data (local, softdep, writes: sync 2 async 0) /dev/wd0g on /var (local, writes: sync 2 async 0) procfs on /proc (local, writes: sync 0 async 0) something is doing syncronous writes to a r/o FS.. who? that's easy.. # mount -u -ordonly / Breakpoint at _mount: pushl %ebp db> b bwrite db> c Breakpoint at _bwrite: pushl %ebp db> tr _bwrite(f245ccf4,0) at _bwrite _ffs_sbupdate(f045f200,1) at _ffs_sbupdate+0xb0 _ffs_mount(f045fc00,efbfd268,efbfd1cc,f3666eb8,f364e740) at _ffs_mount+0x1e5 _mount(f364e740,f3666f94,0,efbfd268,1201) at _mount+0x416 _syscall(27,27,1201,efbfd268,efbfd238) at _syscall+0x187 _Xsyscall() at _Xsyscall+0x35 --- syscall 0x15, eip = 0x6a95, esp = 0xefbfd188, ebp = 0xefbfd238 --- db> apparently _ffs_sbupdate() doesn't know to check if the fs is readonly.. julian If you are doing a downgrade from r/w to r/o then the update is correct. If it is r/o to r/o then it is definitely a bug. ~Kirk
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199802040044.QAA09193>