[PATCH] squashfs broken when pageszie > blocksize

[PATCH] squashfs broken when pageszie > blocksize

[Doug Chapman]

Squashfs is broken on any system where the pageszie is larger than either
the block size of the squashfs image or larger than the metadata size (8192).
This is easily fixed by ensuring cache->pages is always > 0.

Signed-off-by: Doug Chapman <doug.chapman@hp.com>
Cc: Phillip Lougher <phillip@lougher.demon.co.uk>

---
diff --git a/fs/squashfs/cache.c b/fs/squashfs/cache.c
index 1c4739e..c31aa89 100644
--- a/fs/squashfs/cache.c
+++ b/fs/squashfs/cache.c
@@ -252,6 +252,7 @@ struct squashfs_cache *squashfs_cache_init(char *name, int entries,
 	cache->entries = entries;
 	cache->block_size = block_size;
 	cache->pages = block_size >> PAGE_CACHE_SHIFT;
+	cache->pages = cache->pages?cache->pages:1;
 	cache->name = name;
 	cache->num_waiters = 0;
 	spin_lock_init(&cache->lock);

Re: [PATCH] squashfs broken when pageszie > blocksize

[Phillip Lougher]

Doug Chapman wrote:
> Squashfs is broken on any system where the pageszie is larger than either
> the block size of the squashfs image or larger than the metadata size (8192).
> This is easily fixed by ensuring cache->pages is always > 0.
> 
> Signed-off-by: Doug Chapman <doug.chapman@hp.com>
> Cc: Phillip Lougher <phillip@lougher.demon.co.uk>
> 

Good catch.  BTW what is the page size on ia64?

For completeness there should also be a mount time check that
the block size used in the Squashfs filesystem is at least as
large as the pagesize.  This doesn't matter for metadata
(size 8192), but does matter for data blocks, because Squashfs
read_page doesn't handle block_size < pagesize.

Squashfs read_page used to handle block_size < pagesize but I
removed that code when I removed filesystem support for block_sizes
< 4K.  If ia64 users (or any other architecture with larger than
4K pagesizes) want to use Squashfs filesystems with blocksizes less
than their pagesize, then I can put that code back in.

Phillip

Re: [PATCH] squashfs broken when pageszie > blocksize

[Geert Uytterhoeven]

On Mon, 13 Apr 2009, Phillip Lougher wrote:
> Doug Chapman wrote:
> > Squashfs is broken on any system where the pageszie is larger than either
> > the block size of the squashfs image or larger than the metadata size
> > (8192).
> > This is easily fixed by ensuring cache->pages is always > 0.
> > 
> > Signed-off-by: Doug Chapman <doug.chapman@hp.com>
> > Cc: Phillip Lougher <phillip@lougher.demon.co.uk>
> 
> Good catch.  BTW what is the page size on ia64?

A quick grep on PAGE_{SIZE,SHIFT} in asm/page.h shows:

ia64, mips, sh: 4/8/16/64 KiB
parisc: 4/16/64 KiB
alpha, cris: 8 KiB
powerpc: 4/16/64/256 KiB

> For completeness there should also be a mount time check that
> the block size used in the Squashfs filesystem is at least as
> large as the pagesize.  This doesn't matter for metadata
> (size 8192), but does matter for data blocks, because Squashfs
> read_page doesn't handle block_size < pagesize.
> 
> Squashfs read_page used to handle block_size < pagesize but I
> removed that code when I removed filesystem support for block_sizes
> < 4K.  If ia64 users (or any other architecture with larger than
> 4K pagesizes) want to use Squashfs filesystems with blocksizes less
> than their pagesize, then I can put that code back in.

So I think it's a good idea to resurrect that code...

With kind regards,

Geert Uytterhoeven
Software Architect

Sony Techsoft Centre Europe
The Corporate Village · Da Vincilaan 7-D1 · B-1935 Zaventem · Belgium

Phone:    +32 (0)2 700 8453
Fax:      +32 (0)2 700 8622
E-mail:   Geert.Uytterhoeven@sonycom.com
Internet: http://www.sony-europe.com/

A division of Sony Europe (Belgium) N.V.
VAT BE 0413.825.160 · RPR Brussels
Fortis · BIC GEBABEBB · IBAN BE41293037680010

Re: [Squashfs-devel] [PATCH] squashfs broken when pageszie > blocksize

[Phillip Lougher]

Geert Uytterhoeven wrote:

> A quick grep on PAGE_{SIZE,SHIFT} in asm/page.h shows:
> 
> ia64, mips, sh: 4/8/16/64 KiB
> parisc: 4/16/64 KiB
> alpha, cris: 8 KiB
> powerpc: 4/16/64/256 KiB
> 
>> For completeness there should also be a mount time check that
>> the block size used in the Squashfs filesystem is at least as
>> large as the pagesize.  This doesn't matter for metadata
>> (size 8192), but does matter for data blocks, because Squashfs
>> read_page doesn't handle block_size < pagesize.
>>
>> Squashfs read_page used to handle block_size < pagesize but I
>> removed that code when I removed filesystem support for block_sizes
>> < 4K.  If ia64 users (or any other architecture with larger than
>> 4K pagesizes) want to use Squashfs filesystems with blocksizes less
>> than their pagesize, then I can put that code back in.
> 
> So I think it's a good idea to resurrect that code...
> 

Yes, something else to add to my ASAP TODO list.

Phillip