linux-kernel - Re: [PATCH] swiotlb: ensure io_tlb_default

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]

Date:   Mon, 11 Jul 2022 11:07:17 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     Ben Dooks <ben.dooks@...ive.com>, linux-kernel@...r.kernel.org,
        iommu@...ts.linux.dev, iommu@...ts.linux-foundation.org,
        Christoph Hellwig <hch@....de>
Cc:     Sudip Mukherjee <sudip.mukherjee@...ive.com>,
        Jude Onyenegecha <jude.onyenegecha@...ive.com>,
        Marek Szyprowski <m.szyprowski@...sung.com>
Subject: Re: [PATCH] swiotlb: ensure io_tlb_default_mem spinlock always
 initialised

On 2022-07-11 08:26, Ben Dooks wrote:
> On 08/07/2022 21:32, Robin Murphy wrote:
>> On 2022-07-08 18:08, Ben Dooks wrote:
>>> If the io_tlb_default_mem is used by a device that then gets used
>>> by the swiotlb code, the spinlock warning is triggered causing a
>>> lot of stack-trace.
>>
>> Hang on, how have we got as far as trying to allocate out of an 
>> uninitialised SWIOTLB at all? That seems like either something's gone 
>> more fundamentally wrong or we're missing a proper check somewhere. I 
>> don't think papering over it like this is right.
>>
>> Thanks,
>> Robin
> We have a system where we have no DMA capable memory in the 32bit
> window, which means even if we initialise swiotlb we don't have
> any dma capable memory. The code says go use an IOMMU but we don't
> have one of those either (and all peripherals are capable of DMA
> from any of the memory, so there shouldn't be the need for one)
> 
> Is there any other way of fixing this DMA allocation issue?

If none of your peripherals should need SWIOTLB, then the fact that
you're ending up in swiotlb_map() at all is a clear sign that
something's wrong. Most likely someone's forgotten to set their DMA
masks correctly.

However, by inspection it seems we do have a bug here as well, for which
the correct fix should be as below. The fireworks you're *supposed* to
get in that situation are considerably louder and more obvious than a
DEBUG_SPINLOCK complaint ;)

Thanks,
Robin.

----->8-----
Subject: [PATCH] swiotlb: Fail map correctly with failed io_tlb_default_mem

In the failure case of trying to use a buffer which we'd previously
failed to allocate, the "!mem" condition is no longer sufficient since
io_tlb_default_mem became static and assigned by default. Update the
condition to work as intended per the rest of that conversion.

Fixes: 463e862ac63e ("swiotlb: Convert io_default_tlb_mem to static allocation")
Signed-off-by: Robin Murphy <robin.murphy@....com>
---
  kernel/dma/swiotlb.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index cb50f8d38360..5830dce6081b 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -580,7 +580,7 @@ phys_addr_t swiotlb_tbl_map_single(struct device *dev, phys_addr_t orig_addr,
  	int index;
  	phys_addr_t tlb_addr;
  
-	if (!mem)
+	if (!mem || !mem->nslabs)
  		panic("Can not allocate SWIOTLB buffer earlier and can't now provide you with the DMA bounce buffer");
  
  	if (cc_platform_has(CC_ATTR_MEM_ENCRYPT))
-- 
2.36.1.dirty