linux-kernel - Re: [PATCH] firewire: fw-sbp2: enforce s/g segment size limit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <48A2AB8E.1090102@s5r6.in-berlin.de>
Date:	Wed, 13 Aug 2008 11:38:22 +0200
From:	Stefan Richter <stefanr@...6.in-berlin.de>
To:	Grant Grundler <grundler@...gle.com>
CC:	linux1394-devel@...ts.sourceforge.net,
	FUJITA Tomonori <fujita.tomonori@....ntt.co.jp>,
	linux-scsi@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] firewire: fw-sbp2: enforce s/g segment size limit

Grant Grundler wrote:
> On Sat, Aug 9, 2008 at 11:21 AM, Stefan Richter
>> -                       orb->page_table[j].low = cpu_to_be32(sg_addr);
>> -                       orb->page_table[j].high = cpu_to_be32(l << 16);
> 
> I didn't check the rest of the driver - but it would be good if it
> explicitly called dma_set_mask() or  pci_dma_set_mask() with a 32-bit
> mask value. Most drivers assume 32-bit and that's why I point this
> out.

I thought about this and currently think that I should not do this. 
There is the API to set the mask, but there is no API to _decrease_ the 
mask only.  The SBP-2 protocol driver should not set DMA_32BIT_MASK if 
any other driver already set for example DMA_31BIT_MASK for whatever reason.

For now, this is no issue.  FireWire low-level drivers exist only for 
controllers with 32 bit local bus addressing capability.  I am sure that 
somebody will remember to modify the SBP-2 driver(s) if there will ever 
be a controller type and a driver for it which can address more than 4 GB.

On the other hand, we also currently have no reason to set a smaller 
mask.  We recently discovered a chip bug which requires DMA_31BIT_MASK 
for a special mode of data reception (only for cache-coherent DMA 
though), but we now work around this chip bug by simply falling back to 
an alternative mode.
-- 
Stefan Richter
-=====-==--- =--- -==-=
http://arcgraph.de/sr/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/