[lkml]   [2001]   [Dec]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
Subject[PATCH] cciss 2.5.1 for 2.5.1
I wrote:
> [...]
> > Here is a new patch against 2.5.1-pre11:
> >
> > Having played with it some more, I see it is seriously flawed
> in the area of SCSI tape support and locking. [...]

Ok. I think I fixed all that. (Main problem was how I was allocating
my commands for SCSI side of the driver, this bug has been there a
long time, I'm surprised it lasted as long as it did without me seeing it.)

I tested it with 2 tape drives going simultaneously
on a 2 processor system, which was embarassingly
fatal for the previous.drivers.

About the prior patch,
had some complaint that first arg of pci_*_consistent was NULL in
some places in the driver. So I included comments explaining why
it is that way. (Documentation/DMA-mapping.txt says this is

Another complaint that various explicit initializations
to NULL or 0 should be left up to the compiler. Whatever.
Change them if you must.

An excerpt from the patch:
Thu Dec 20 09:43:04 CST 2001
v. 2.5.1 * Fixed longstanding problem in allocation of SCSI commands
which meant physical addresses of cmds for tape drives could
be wrong. Problem showed easily if 2 tape drives used
simultaneously, though in theory it could show with only 1
tape drive.
* Fixed races/deadlocks resulting from first attempt
at removing io_request_lock
* Changed pci_alloc_consistent call in cciss_scsi.c so
that the SCSI command pool would be guaranteed to have
addresses that fit through the 32 bit command register.
Thu Dec 13 16:39:05 CST 2001
v. 2.5.0 * Removal of io_request_lock for 2.5.x kernels.
* no longer sets block_size to zero for volumes where
read capacity fails (caused div by zero)
* reset MAXSGENTRIES back to 31 (controllers reject 32)
Thu Oct 4 14:29:15 CDT 2001 Changelog begins.
v. 2.4.21 * Added support for SCSI tape drives (Steve Cameron)
* Added support for dynamically adding and removing
logical volumes, This implies that the disk index
("x" in /dev/cciss/c*b*dx*) is no longer a contiguous
sequential series (e.g. 0,1,2,3) as it now maps to logical
volume numbers directly and thus there may be gaps.
(Charles White)
-- steve

Do You Yahoo!?
Check out Yahoo! Shopping and Yahoo! Auctions for all of
your unique holiday gifts! Buy at
or bid at
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:18    [W:0.035 / U:1.088 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site