lkml.org 
[lkml]   [2007]   [Jun]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Oops in a driver while using SLUB as a SLAB allocator
On Mon, 25 Jun 2007, Hugh Dickins wrote:

> > In many situations the page struct passed to flush_dcache_page is
> > simply used to calculate the virtual address. So its mostly harmless.
> > Trouble starts when page attributes like the mapping is used.
>
> Mostly harmless indeed. I don't understand why you insist on trying
> to complicate the situation. flush_dcache_page is only expected to
> do something on pages mapped into userspace (correct me if I'm wrong
> there), it's expected to do nothing on kmalloc'ed pages. It's
> been working that way for years, and will continue to work that way
> with slub, providing either page_mapping or flush_dcache_page checks
> PageSlab to avoid oopsing on page->mapping.

It is definitely intended to work. Otherwise we would not have code
like this:

christoph@fly:~/linux-2.6$ find . -name "*.c" | xargs grep "flush_dcache_page"|grep virt
./drivers/scsi/scsi_tgt_if.c: flush_dcache_page(virt_to_page(ev));
./drivers/scsi/scsi_tgt_if.c: flush_dcache_page(virt_to_page(ev));


> 2.6.22-rc6 has page_mapping making that check: we could argue about
> which is the better site for it, there are good arguments both ways
> (page_mapping is the correct place, flush_dcache_page is the more
> efficient place), I suggest we leave it as is.

Ok. I think your patch is fine as a quick fix for 2.6.22. I am a bit
uneasy with that given that its in such a broadly used function while its
only use is to enable flush_dcache_page to work. But we need the general
issue taken care of after 2.6.22.

> > A kmalloc slab object (even 64 byte) may be crossing a page boundary
> > with a ARCH_KMALLOC_MINALIGN of 4 or 8. So I think that
> > flush_dcache_range *must* be used rather than flush_dcache_page.
>
> Why???? All we require of flush_dcache_page is that it not oops on
> the first page in the range: we don't need to change over to
> flush_dcache_range for that.

As explained about: There are corner cases in which it does not work. You
seem to assume that flush_dcache_page can become a no op. That may not be
true on platforms that need explicit cache flushing for a DMA engine to
access a data structure. The above listed use suggests that the caller
expects flushing to occur correctly.



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2007-06-25 19:03    [W:0.149 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site