From: James Bottomley <James.Bottomley@HansenPartnership.com> Date: 2013-01-16 10:32:43
On Wed, 2013-01-16 at 12:07 +0530, Subhash Jadavani wrote:
Now consider this call stack from MMC block driver (this is on the ARmv7
based board):
[ 98.918174] [<c001b50c>] (v7_dma_inv_range+0x30/0x48) from
[<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c)
[ 98.927819] [<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c) from
[<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c)
[ 98.937982] [<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c) from
[<c0017ff8>] (dma_map_sg+0x3c/0x114)
OK, so this is showing that ARM itself is making the assumption that the
pages are contiguous in the page offset map.
Fix this by doing the increment via the pfn, which will do the right
thing whatever the memory model.
Signed-off-by: James Bottomley <redacted>
---
On Wed, 2013-01-16 at 12:07 +0530, Subhash Jadavani wrote:
quoted
Now consider this call stack from MMC block driver (this is on the ARmv7
based board):
[ 98.918174] [<c001b50c>] (v7_dma_inv_range+0x30/0x48) from
[<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c)
[ 98.927819] [<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c) from
[<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c)
[ 98.937982] [<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c) from
[<c0017ff8>] (dma_map_sg+0x3c/0x114)
OK, so this is showing that ARM itself is making the assumption that the
pages are contiguous in the page offset map.
Fix this by doing the increment via the pfn, which will do the right
thing whatever the memory model.
Signed-off-by: James Bottomley <redacted>
Thanks James. Yes, it make sense to fix the ARM code itself if it is the
only one giving this trouble.
I have tried your change below and it also fixes this issue (without
having my blk-merge patch). I will forward your change to Russel King to
see what he thinks about it.
Regards,
Subhash
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Russell,
Is it possible to pick up James patch below? Thread here:
http://comments.gmane.org/gmane.linux.kernel.mmc/18670, have the details
on the motivation behind this fix.
Regards,
Subhash
-------- Original Message --------
Subject: Re: [PATCH v2 1/1] block: blk-merge: don't merge the pages
with non-contiguous descriptors
Date: Wed, 16 Jan 2013 18:09:14 +0530
From: Subhash Jadavani <redacted>
To: James Bottomley <James.Bottomley@HansenPartnership.com>
CC: linux-kernel at vger.kernel.org, linux-scsi at vger.kernel.org,
linux-mmc at vger.kernel.org, linux-arm-msm at vger.kernel.org,
martin.petersen at oracle.com, asias at redhat.com, tj at kernel.org,
linux-arm-kernel at lists.infradead.org, Russell King [off-list ref]
On 1/16/2013 4:02 PM, James Bottomley wrote:
On Wed, 2013-01-16 at 12:07 +0530, Subhash Jadavani wrote:
quoted
Now consider this call stack from MMC block driver (this is on the ARmv7
based board):
[ 98.918174] [<c001b50c>] (v7_dma_inv_range+0x30/0x48) from
[<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c)
[ 98.927819] [<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c) from
[<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c)
[ 98.937982] [<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c) from
[<c0017ff8>] (dma_map_sg+0x3c/0x114)
OK, so this is showing that ARM itself is making the assumption that the
pages are contiguous in the page offset map.
Fix this by doing the increment via the pfn, which will do the right
thing whatever the memory model.
Signed-off-by: James Bottomley <redacted>
Thanks James. Yes, it make sense to fix the ARM code itself if it is the
only one giving this trouble.
I have tried your change below and it also fixes this issue (without
having my blk-merge patch). I will forward your change to Russel King to
see what he thinks about it.
Regards,
Subhash
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Someone should also audit the arm kernel code for more of these linear
page array assumptions. I'm guessing that when sparsemem was added to
arm over a year ago, whoever did it either didn't audit or missed a few.
James
Someone should also audit the arm kernel code for more of these linear
page array assumptions. I'm guessing that when sparsemem was added to
arm over a year ago, whoever did it either didn't audit or missed a few.
No, that's a bad assumption. We've had discontigmem for years - maybe
something like 12 years. I switched everything over to sparsemem, and
sparsemem has been used on ARM for years too:
commit 05944d74bc28fffbcce159cb915d0acff82f30a1
Author: Russell King [off-list ref]
Date: Thu Nov 30 20:43:51 2006 +0000
[ARM] Add initial sparsemem support
Signed-off-by: Russell King [off-list ref]
However, there's a big problem with this: very few of the lead people
have machines which suffer from this disability, so there's very little
testing of it - and there's very little testing of new code with it.
The patch which originally introduced this code which your patch
touches was part of adding highmem support to ARM:
commit 43377453af83b8ff8c1c731da1508bd6b84ebfea
Author: Nicolas Pitre [off-list ref]
Date: Thu Mar 12 22:52:09 2009 -0400
[ARM] introduce dma_cache_maint_page()
This is a helper to be used by the DMA mapping API to handle cache
maintenance for memory identified by a page structure instead of a
virtual address. Those pages may or may not be highmem pages, and
when they're highmem pages, they may or may not be virtually mapped.
When they're not mapped then there is no L1 cache to worry about. But
even in that case the L2 cache must be processed since unmapped highmem
pages can still be L2 cached.
Signed-off-by: Nicolas Pitre [off-list ref]
some three years later, and has been through a number of revisions since.
I'd really like to get rid of sparsemem so it's one less failure case, but
alas there's a relatively small bunch of folk who rely upon it. That
means it's always going to be more buggy.
Someone should also audit the arm kernel code for more of these linear
page array assumptions. I'm guessing that when sparsemem was added to
arm over a year ago, whoever did it either didn't audit or missed a few.
No, that's a bad assumption. We've had discontigmem for years - maybe
something like 12 years.
Discontigmem doesn't suffer from this particular problem: The breaks in
the page arrays in a discontigmem environment represent real breaks in
the physical memory map, so there can never arise a situation where
page++ wouldn't get you the next valid page, provided there's still a
real contiguous page of physical memory present in the system, of
course.
Sparsemem tried to be cleverer and it broke that assumption. I'm
starting to wonder if there are other places in the kernel which make it
and get broken by sparsemem.
I have patches for parisc to switch it over to sparsemem from
discontigmem ... I haven't pushed them yet, fortunately, and I certainly
didn't check our underlying assumptions for this problem.
I switched everything over to sparsemem, and
sparsemem has been used on ARM for years too:
commit 05944d74bc28fffbcce159cb915d0acff82f30a1
Author: Russell King [off-list ref]
Date: Thu Nov 30 20:43:51 2006 +0000
[ARM] Add initial sparsemem support
Signed-off-by: Russell King [off-list ref]
However, there's a big problem with this: very few of the lead people
have machines which suffer from this disability, so there's very little
testing of it - and there's very little testing of new code with it.
Anyone can test this just by enabling sparsemem with an artificially
small SECTION_SIZE_BITS (that makes it much more likely you'll hit a
boundary).
The patch which originally introduced this code which your patch
touches was part of adding highmem support to ARM:
commit 43377453af83b8ff8c1c731da1508bd6b84ebfea
Author: Nicolas Pitre [off-list ref]
Date: Thu Mar 12 22:52:09 2009 -0400
[ARM] introduce dma_cache_maint_page()
This is a helper to be used by the DMA mapping API to handle cache
maintenance for memory identified by a page structure instead of a
virtual address. Those pages may or may not be highmem pages, and
when they're highmem pages, they may or may not be virtually mapped.
When they're not mapped then there is no L1 cache to worry about. But
even in that case the L2 cache must be processed since unmapped highmem
pages can still be L2 cached.
Signed-off-by: Nicolas Pitre [off-list ref]
some three years later, and has been through a number of revisions since.
I'd really like to get rid of sparsemem so it's one less failure case, but
alas there's a relatively small bunch of folk who rely upon it. That
means it's always going to be more buggy.
Heh, I'd like to do that too on parisc; the problem is that the memory
holes in our platform are just to big to use a linear array on. And now
that the slub people want DISCONTIGMEM deprecated, there's not much
choice.
James
From: Russell King - ARM Linux <hidden> Date: 2013-01-16 23:14:27
On Wed, Jan 16, 2013 at 10:32:35AM +0000, James Bottomley wrote:
On Wed, 2013-01-16 at 12:07 +0530, Subhash Jadavani wrote:
quoted
Now consider this call stack from MMC block driver (this is on the ARmv7
based board):
[ 98.918174] [<c001b50c>] (v7_dma_inv_range+0x30/0x48) from
[<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c)
[ 98.927819] [<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c) from
[<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c)
[ 98.937982] [<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c) from
[<c0017ff8>] (dma_map_sg+0x3c/0x114)
OK, so this is showing that ARM itself is making the assumption that the
pages are contiguous in the page offset map.
Fix this by doing the increment via the pfn, which will do the right
thing whatever the memory model.
Signed-off-by: James Bottomley <redacted>
Ok. What would you like the patch summary line for this to be -
the existing one seems to be a little wrong given the content of
this patch...
From: James Bottomley <James.Bottomley@HansenPartnership.com> Date: 2013-01-17 08:54:55
On Wed, 2013-01-16 at 23:14 +0000, Russell King - ARM Linux wrote:
On Wed, Jan 16, 2013 at 10:32:35AM +0000, James Bottomley wrote:
quoted
On Wed, 2013-01-16 at 12:07 +0530, Subhash Jadavani wrote:
quoted
Now consider this call stack from MMC block driver (this is on the ARmv7
based board):
[ 98.918174] [<c001b50c>] (v7_dma_inv_range+0x30/0x48) from
[<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c)
[ 98.927819] [<c0017b8c>] (dma_cache_maint_page+0x1c4/0x24c) from
[<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c)
[ 98.937982] [<c0017c28>] (___dma_page_cpu_to_dev+0x14/0x1c) from
[<c0017ff8>] (dma_map_sg+0x3c/0x114)
OK, so this is showing that ARM itself is making the assumption that the
pages are contiguous in the page offset map.
Fix this by doing the increment via the pfn, which will do the right
thing whatever the memory model.
Signed-off-by: James Bottomley <redacted>
Ok. What would you like the patch summary line for this to be -
the existing one seems to be a little wrong given the content of
this patch...
how about
arm: fix struct page iterator in dma_cache_maint() to work with
sparsemem
?
James
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Probably page = nth_page(page, 1) is the better form.
It's the same thing.
I'd actually prefer page = pfn_to_page(page_to_pfn(page) + 1); because
it makes the code look like the hack it is. The preferred form for all
iterators like this should be to iterate over the pfn instead of a
pointer into the page arrays, because that will always work correctly no
matter how many weird and wonderful memory schemes we come up with.
James
Probably page = nth_page(page, 1) is the better form.
It's the same thing.
I'd actually prefer page = pfn_to_page(page_to_pfn(page) + 1); because
it makes the code look like the hack it is. The preferred form for all
iterators like this should be to iterate over the pfn instead of a
pointer into the page arrays, because that will always work correctly no
matter how many weird and wonderful memory schemes we come up with.
From: Russell King - ARM Linux <hidden> Date: 2013-01-17 10:48:03
On Thu, Jan 17, 2013 at 10:37:42AM +0000, Russell King - ARM Linux wrote:
On Thu, Jan 17, 2013 at 09:11:20AM +0000, James Bottomley wrote:
quoted
I'd actually prefer page = pfn_to_page(page_to_pfn(page) + 1); because
it makes the code look like the hack it is. The preferred form for all
iterators like this should be to iterate over the pfn instead of a
pointer into the page arrays, because that will always work correctly no
matter how many weird and wonderful memory schemes we come up with.
So, why don't we update the code to do that then?
Also, couldn't the addition of the scatterlist offset to the page also
be buggy too?
So, what about this patch which addresses both additions by keeping our
iterator as a pfn as you suggest. It also simplifies some of the code
in the loop too.
Can the original folk with the problem test this patch?
arch/arm/mm/dma-mapping.c | 18 ++++++++++--------
1 files changed, 10 insertions(+), 8 deletions(-)
From: James Bottomley <James.Bottomley@HansenPartnership.com> Date: 2013-01-17 11:01:54
On Thu, 2013-01-17 at 10:47 +0000, Russell King - ARM Linux wrote:
On Thu, Jan 17, 2013 at 10:37:42AM +0000, Russell King - ARM Linux wrote:
quoted
On Thu, Jan 17, 2013 at 09:11:20AM +0000, James Bottomley wrote:
quoted
I'd actually prefer page = pfn_to_page(page_to_pfn(page) + 1); because
it makes the code look like the hack it is. The preferred form for all
iterators like this should be to iterate over the pfn instead of a
pointer into the page arrays, because that will always work correctly no
matter how many weird and wonderful memory schemes we come up with.
So, why don't we update the code to do that then?
We can, but it involves quite a rewrite within the arm dma-mapping code
to use pfn instead of page. It looks like it would make the code
cleaner because there are a lot of page_to_pfn transformations in there.
However, the current patch is the simplest one for stable and I don't
actually have any arm build and test environments.
Also, couldn't the addition of the scatterlist offset to the page also
be buggy too?
No, fortunately, offset must be within the first page from the point of
view of block generated sg lists. As long as nothing within arm
violates this, it should be a safe assumption ... although the code
seems to assume otherwise.
James
quoted hunk
So, what about this patch which addresses both additions by keeping our
iterator as a pfn as you suggest. It also simplifies some of the code
in the loop too.
Can the original folk with the problem test this patch?
arch/arm/mm/dma-mapping.c | 18 ++++++++++--------
1 files changed, 10 insertions(+), 8 deletions(-)
From: Russell King - ARM Linux <hidden> Date: 2013-01-17 11:04:59
On Thu, Jan 17, 2013 at 11:01:47AM +0000, James Bottomley wrote:
On Thu, 2013-01-17 at 10:47 +0000, Russell King - ARM Linux wrote:
quoted
Also, couldn't the addition of the scatterlist offset to the page also
be buggy too?
No, fortunately, offset must be within the first page from the point of
view of block generated sg lists. As long as nothing within arm
violates this, it should be a safe assumption ... although the code
seems to assume otherwise.
Are you absolutely sure about that? I believe I have seen cases where
that has been violated in the past, though it was many years ago.
From: James Bottomley <James.Bottomley@HansenPartnership.com> Date: 2013-01-17 11:19:27
On Thu, 2013-01-17 at 11:04 +0000, Russell King - ARM Linux wrote:
On Thu, Jan 17, 2013 at 11:01:47AM +0000, James Bottomley wrote:
quoted
On Thu, 2013-01-17 at 10:47 +0000, Russell King - ARM Linux wrote:
quoted
Also, couldn't the addition of the scatterlist offset to the page also
be buggy too?
No, fortunately, offset must be within the first page from the point of
view of block generated sg lists. As long as nothing within arm
violates this, it should be a safe assumption ... although the code
seems to assume otherwise.
Are you absolutely sure about that? I believe I have seen cases where
that has been violated in the past, though it was many years ago.
From the point of view of the block layer, absolutely: the scatterlist
is generated from an array of bio_vecs. Each bio_vec is a page, offset
and length element and obeys the rule that offset must be within the
page and offset + length cannot stray over the page.
From the point of view of other arm stuff, I don't know.
From: Russell King - ARM Linux <hidden> Date: 2013-01-17 11:40:57
On Thu, Jan 17, 2013 at 11:19:21AM +0000, James Bottomley wrote:
On Thu, 2013-01-17 at 11:04 +0000, Russell King - ARM Linux wrote:
quoted
On Thu, Jan 17, 2013 at 11:01:47AM +0000, James Bottomley wrote:
quoted
On Thu, 2013-01-17 at 10:47 +0000, Russell King - ARM Linux wrote:
quoted
Also, couldn't the addition of the scatterlist offset to the page also
be buggy too?
No, fortunately, offset must be within the first page from the point of
view of block generated sg lists. As long as nothing within arm
violates this, it should be a safe assumption ... although the code
seems to assume otherwise.
Are you absolutely sure about that? I believe I have seen cases where
that has been violated in the past, though it was many years ago.
quoted
From the point of view of the block layer, absolutely: the scatterlist
is generated from an array of bio_vecs. Each bio_vec is a page, offset
and length element and obeys the rule that offset must be within the
page and offset + length cannot stray over the page.
Well, I found it when working on the mmc stuff initially, long before
it got complex. The scatterlists were unmodified from the block layer,
and I'm positive I saw occasions where the offset in the scatter lists
were larger than PAGE_SIZE.
quoted
From the point of view of other arm stuff, I don't know.
On 1/17/2013 4:17 PM, Russell King - ARM Linux wrote:
On Thu, Jan 17, 2013 at 10:37:42AM +0000, Russell King - ARM Linux wrote:
quoted
On Thu, Jan 17, 2013 at 09:11:20AM +0000, James Bottomley wrote:
quoted
I'd actually prefer page = pfn_to_page(page_to_pfn(page) + 1); because
it makes the code look like the hack it is. The preferred form for all
iterators like this should be to iterate over the pfn instead of a
pointer into the page arrays, because that will always work correctly no
matter how many weird and wonderful memory schemes we come up with.
So, why don't we update the code to do that then?
Also, couldn't the addition of the scatterlist offset to the page also
be buggy too?
So, what about this patch which addresses both additions by keeping our
iterator as a pfn as you suggest. It also simplifies some of the code
in the loop too.
Can the original folk with the problem test this patch?
Yes, this patch also fixes the issue. You may add: Tested-by: Subhash
Jadavani [off-list ref] .
Regards,
Subhash
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html