Trouble with APM suspend in 5.3-R

[M. Warner Losh]

In message: <41F495FE.6000907 at orel.ru>
            Andrew Belashov <bel at orel.ru> writes:
: -----BEGIN PGP SIGNED MESSAGE-----
: Hash: SHA1
: 
: Hello, Warner!
: 
: M. Warner Losh wrote:
: | In message: <41F02CE1.5090207 at acm.org>
: |             Dave Walton <dwalton at acm.org> writes:
: | : Andrew Belashov wrote:
: | : > Dave Walton wrote:
: | : >
: | : >> I've been unable to get APM suspend/resume to work with 5.3-R on my
: | : >> Thinkpad 770Z.  As released, 'apm -z' causes a lockup (it worked fine
: | : >> in 4.x).  Revision 1.233 of ata-all.c fixes that, and 'apm -z' now
: | : >> causes the system to properly suspend to disk, as it had before.
: | : >> However, when the system resumes, it spits out three errors regarding
: | : >> pir0, then panics.  This happens with or without your patch applied to
: | : >> ata-all.c.
: | : >>
: | : >
: | : > Try attached patch as workaround.
: | :
: | : A good effort!  That is precisely the error I saw.  Hopefully, that
: | : removed call to pci_pir_biosroute() doesn't do anything important.
: | :
: | : Unfortunately, this had no effect at all on the panic.  Please see my
: | : next reply to Gleb Smirnoff for details on the panic.  Perhaps it will
: | : mean something to you.
: |
: | Chances are the right fix is to try the route, but ignore errors...
: 
: I agree. But my old notebook recursively goes in cycles by calling
: bios32(PCIBIOS_ROUTE_INTERRUPT) after resume. As result: kernel stack
: overflow, double panic.
: 
: BIOS update is not available. No known problems in Windows 2000 Pro.
: 
: After removing call to pci_pir_biosroute(), suspend/resume works completely
: on my notebook...

This sounds like you disagree with me.  I'm saying that one should
call pci_pir_biosroute() anyway, and then ignore the return code.  It
sounds like you are saying that this call is fatal in a resume
context..

Warner