Bug 572004

Adam,
some of the scsi functions (scsi_dma_map,..) were backported to RHEL5 in the past, you can use them next time if you wish.

Patch posted today on rhkl.

Adam,
thi came to attention during internal review:
+/* This function is called on PCI resume */
+static int twa_resume(struct pci_dev *pdev)
+{
+	int retval = 0;
...
+out_disable_device:
+	scsi_remove_host(host);
+	pci_disable_device(pdev);
+
+	return retval;
+} /* End twa_resume() */
+#endif

In this failure path, are you supposed to remove the scsi host? If you 
rmmod the driver then twa_remove will not get called will it?
If you are supposed to remove it then  you need a scsi_host_put() call 
(to release the last refcount).

Tomas,

Are you talking about the case where we are not the boot controller but
resume fails (and the system hasn't crashed), then someone wants to rmmod the driver? 

What is your suggested fix for this?

Remove scsi_remove_host() call from twl_resume() ?

or:

Replace scsi_remove_host() with scsi_host_put() ?

Looking through the latest kernel 2.6.36-rc2, there seems to be some 
inconsistencies as to what scsi drivers do in this case.  

Some drivers (such as megaraid_sas.c/pmcraid.c) are doing this:

scsi_host_put()
pci_disable_device()

While others do nothing, except return non zero from hostt->resume().

-Adam

(In reply to comment #3)
> Are you talking about the case where we are not the boot controller but
> resume fails (and the system hasn't crashed), then someone wants to rmmod the
> driver? 
Yes, exactly.

> What is your suggested fix for this?
> 
> Remove scsi_remove_host() call from twl_resume() ?
> 
> or:
> 
> Replace scsi_remove_host() with scsi_host_put() ?
> 
> Looking through the latest kernel 2.6.36-rc2, there seems to be some 
> inconsistencies as to what scsi drivers do in this case.
> 
> Some drivers (such as megaraid_sas.c/pmcraid.c) are doing this:
> 
> scsi_host_put()
> pci_disable_device()

I think that when driver fails to resume that that when rmmod the remove function is still called. That for me means the driver usually  should call scsi_remove_host + scsi_host_put only in the remove function. So I'd prefer the latest option: (with a small note - I'm by far not an scsi layer guru)
> While others do nothing, except return non zero from hostt->resume().

Adam,
have you come to a solution for this, or do you have an other idea?

Nope, I think we just need a 1 line patch that removes the
scsi_remote_host() call from twa_resume().   Should I attach one to this
bugzilla?

-Adam

(In reply to comment #6)
> Nope, I think we just need a 1 line patch that removes the
> scsi_remote_host() call from twa_resume().   Should I attach one to this
> bugzilla?

Not needed, but please post the correction upstream. 
Btw. I think I have seen this also in other 3w* driver.

in kernel-2.6.18-222.el5
You can download this test kernel from http://people.redhat.com/jwilson/el5

Detailed testing feedback is always welcomed.

~~ Attention Customers and Partners - RHEL 5.6 Public Beta is now available on RHN ~~

A fix for this 'OtherQA' BZ should be present and testable in the release. 

If this Bugzilla is verified as resolved, please update the Verified field above with an appropriate value and include a summary of the testing executed and the results obtained.

If you encounter any issues or have questions while testing, please describe them and set this bug into NEED_INFO. 

If you encounter new defects or have additional patches to request for inclusion, promptly escalate the new issues through your support representative.

Finally, future Beta kernels can be found here:
 http://people.redhat.com/jwilson/el5/

Note: Bugs with the 'OtherQA' keyword require Third-Party testing to confirm the request has been properly addressed. See: https://bugzilla.redhat.com/describekeywords.cgi#OtherQA ).

Redhat,

I have verified this bugzilla with RHEL5.6 (snapshot 2) kernel 2.6.18-232.el5.
I was able to successfully access storage devices using the 3w-9xxx driver, and
the 3w-9xxx driver loaded successfully during OS installation.

-Adam

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2011-0017.html