Re: [PATCH] pci-sysfs: backport fix for 2.6.11.12

[Greg KH <gregkh@suse.de>]

On Tue, May 31, 2005 at 11:57:29PM +0300, Michael S. Tsirkin wrote:
> Quoting r. Greg KH <gregkh@suse.de>:
> > Subject: Re: [PATCH] pci-sysfs: backport fix for 2.6.11.12
> > 
> > On Tue, May 31, 2005 at 07:36:19PM +0300, Michael S. Tsirkin wrote:
> > > Greg, before 2.6.12, pci_write_config in pci-sysfs.c was broken, causing
> > > incorrect data being written to the configuration register,
> > > which in the case of my userspace driver results in system failure.
> > > 
> > > This has been fixed in 2.6.12-rc5:
> > > 
> > > http://www.kernel.org/diff/diffview.cgi?file=%2Fpub%2Flinux%2Fkernel%2Fv2.6%2Ftesting%2Fpatch-2.6.12-rc5.bz2;z=2656
> > > 
> > > Would you please consider merging the fix for 2.6.11.12 as well?
> > 
> > Would you care to split out only the proper part for that fix?  There
> > are a few different patches in that link above.
> 
> Hmm. There's also a new attribute there, that shall wait for 2.6.12.

Yes, please remember, different patches can touch the same file :)

> There was a general cleanup coverting *all* direct uses of char* to
> first cast to u8 *, not only in pci_write_config where it triggers
> an actual bug. Do you want all that cleanup in?
> I can do it, just let me know.

Well, I don't think that just applying one part of the whole main patch
is a good idea at all.  Either do it all or not at all.  The original
patch can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=4c0619add8c3a8b28e7fae8b15cc7b62de2f8148

Along with the proper description and signed-off-by lines.

> > > Alternatively (since there were multiple other changes in pci-sysfs.c), here's
> > > a small patch to fix just this issue.
> > 
> > I don't think this fixes the problem properly.  Can you verify it?
> 
> I did verify it before sending :), my patch below does fix the problem.

But it doesn't fix the main problem, right?

Honestly, this problem has been around for so long, with no real
complaints from anyone (all the distros already patched this fix a long
time ago), and it's too big for the -stable rules, that I do not think
it should go in.

So no, if you really need this fix, use 2.6.12, or a disto kernel, your
fix is not correct.

thanks,

greg k-h