failure in do_get_write_access() at transaction.c:731: "(((jh2bh(jh))->b_state & (1UL << BH_Uptodate)) != 0)"

Christian evilninja at
Mon Aug 14 21:13:06 UTC 2006

On Mon, 14 Aug 2006, Zhao, Joe wrote:
> The problem is not easily reproducible but I have noticed that
> somebody had raised the same issue before
> (

thanks for the link, that really should clear up some things:

Andrew Morton replied in [0] and pointed to a ext3-cvs.patch. 
Unfortunately it's not there any more: try to contact him where this 
file has moved to or try the wayback-machine [1] when it's back online.

Another reply [2] from Stephen Tweedie explained this in depth, 
mentioning 2 fixes:

1) a missing "goto repeat", this should be this patch:

2) something involing parallel blockdevice IO/fs IO. However, I could
    not spot the right patch on the ext3 CVS commit list [3]. Maybe
    the ext3 developers remember...

That, or someone with the lost ext3-cvs.patch should do it.

> I know that the kernel version I am currently running is too old.
> However, since it is running on the legacy product, it is not easy
> to convince the customer to do the upgrade.

4 years of begging to update their product to the present 
state-of-technology, wow ;)

> If you can point me to where the fix is, I will probably just get
> the fix and replace the file(s) while keeping running 2.4.19 kernel.

Although Andrews patch was in ../2.4/2.4.19-pre8/ext3-cvs.patch, he 

> Stephen has a fix in ext3 CVS for this.  I've been playing with that
> fix in 2.5.x.  I guess we'll slot it into 2.4.20-pre.

would it be possible to "upgrade" to 2.4.20?


BOFH excuse #70:

nesting roaches shorted out the ether cable

More information about the Ext3-users mailing list