possible SMP problem in fs/inode.c?

[Tim Walberg]

Posting again, since I never heard anything back the first time,
and I lost access to the e-mail address from which it was posted
(unexpected company layoffs), and I've been having problems posting
to the list due to ORBS (think I've got that fixed now, though, by
bypassing my ISP's relays...). Anyway, don't roast me too bad if
multiple copies get posted somehow - I'm still not quite sure I've
got all the SMTP bumps worked out after a major rework of my
sendmail configuration.


			tw


On 01/11/2000 09:40 -0600, Tim Walberg wrote:
>>	The last few days, I've been getting a bunch of messages
>>	like:
>>	
>>	Jan 11 07:37:21 calvin kernel: iput: inode 09:00/14189 still has locks!
>>	Jan 11 07:37:23 calvin kernel: iput: inode 09:03/996435 still has locks!
>>	Jan 11 07:37:36 calvin kernel: iput: inode 09:03/1195010 still has locks!
>>	Jan 11 07:37:38 calvin kernel: iput: inode 09:03/541708 still has locks!
>>	Jan 11 07:37:38 calvin kernel: iput: inode 00:01/181600258 still has locks!
>>	Jan 11 07:37:40 calvin kernel: iput: inode 09:03/1029138 still has locks!
>>	
>>	I've looked at a number of the files that these represent, and
>>	they are not files that are likely to be open and locked, so my
>>	guess is that, since this is an SMP machine, maybe iput is getting
>>	called on both CPUs with the same inode on (relatively rare)
>>	occasions? I'm not sure that makes sense, though... I've only
>>	seen the messages during periods of high atime activity (i.e.
>>	running find or backups), and I'm not sure this is indicative
>>	of a serious problem, but, since the code generating this is
>>	inside an '#ifdef INODE_PARANOIA', it kinda makes me wonder...
>>	
>>	System is a dual PPro 200MHz using a SuperMicro P6DNE, 512MB,
>>	2 WDC AC418000D 18GB IDE drives using RAID1 for all filesystems.
>>	Kernel is 2.2.13ac3 plus the ide-tape patch for an OnStream ADR
>>	drive.
>>	
>>	Also, I find the messages for inodes on dev 00:01 particularly
>>	curious, since I have no /dev node with those numbers, and
>>	linux/Documentation/devices.txt claims that major 0 is for unnamed
>>	devices. Is this /proc or /dev/pts maybe? Is updating the atime
>>	on such a pseudo-fs useful? Maybe these should be mounted ro...
>>	
>>	Anyway, any insight you can provide me would be useful. Thanks.
>>	
>>	
>>	                                tw
>>	
End of included message

-- 
+--------------------------+------------------------------+
| Tim Walberg              | tewalberg en mediaone.net       |
| 828 Marshall Ct.         | www.concentric.net/~twalberg |
| Palatine, IL 60074       |                              |
+--------------------------+------------------------------+
------------ próxima parte ------------
A non-text attachment was scrubbed...
Name: no disponible
Type: application/pgp-signature
Size: 248 bytes
Desc: no disponible
URL: <https://lists.srvr.mx/pipermail/ayuda/attachments/20000120/8f4b54cf/attachment.asc>