* FAILED: patch "[PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset" failed to apply to 5.12-stable tree
@ 2021-06-25 9:26 gregkh
[not found] ` <CGME20210628065823epcas2p19305f8b888a7fc0e883ec51db61e3bae@epcas2p1.samsung.com>
0 siblings, 1 reply; 15+ messages in thread
From: gregkh @ 2021-06-25 9:26 UTC (permalink / raw)
To: bumyong.lee, chanho61.park, dominique.martinet, hch,
horia.geanta, konrad.wilk
Cc: stable
The patch below does not apply to the 5.12-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 5f89468e2f060031cd89fd4287298e0eaf246bf6 Mon Sep 17 00:00:00 2001
From: Bumyong Lee <bumyong.lee@samsung.com>
Date: Mon, 10 May 2021 18:10:04 +0900
Subject: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
in case of driver wants to sync part of ranges with offset,
swiotlb_tbl_sync_single() copies from orig_addr base to tlb_addr with
offset and ends up with data mismatch.
It was removed from
"swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single",
but said logic has to be added back in.
From Linus's email:
"That commit which the removed the offset calculation entirely, because the old
(unsigned long)tlb_addr & (IO_TLB_SIZE - 1)
was wrong, but instead of removing it, I think it should have just
fixed it to be
(tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
instead. That way the slot offset always matches the slot index calculation."
(Unfortunatly that broke NVMe).
The use-case that drivers are hitting is as follow:
1. Get dma_addr_t from dma_map_single()
dma_addr_t tlb_addr = dma_map_single(dev, vaddr, vsize, DMA_TO_DEVICE);
|<---------------vsize------------->|
+-----------------------------------+
| | original buffer
+-----------------------------------+
vaddr
swiotlb_align_offset
|<----->|<---------------vsize------------->|
+-------+-----------------------------------+
| | | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
2. Do something
3. Sync dma_addr_t through dma_sync_single_for_device(..)
dma_sync_single_for_device(dev, tlb_addr + offset, size, DMA_TO_DEVICE);
Error case.
Copy data to original buffer but it is from base addr (instead of
base addr + offset) in original buffer:
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- size ->|
+-----------------------------------+
|##########| | original buffer
+-----------------------------------+
vaddr
The fix is to copy the data to the original buffer and take into
account the offset, like so:
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- offset ->|<- size ->|
+-----------------------------------+
| |##########| | original buffer
+-----------------------------------+
vaddr
[One fix which was Linus's that made more sense to as it created a
symmetry would break NVMe. The reason for that is the:
unsigned int offset = (tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
would come up with the proper offset, but it would lose the
alignment (which this patch contains).]
Fixes: 16fc3cef33a0 ("swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single")
Signed-off-by: Bumyong Lee <bumyong.lee@samsung.com>
Signed-off-by: Chanho Park <chanho61.park@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: Dominique MARTINET <dominique.martinet@atmark-techno.com>
Reported-by: Horia Geantă <horia.geanta@nxp.com>
Tested-by: Horia Geantă <horia.geanta@nxp.com>
CC: stable@vger.kernel.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 8ca7d505d61c..e50df8d8f87e 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -334,6 +334,14 @@ void __init swiotlb_exit(void)
io_tlb_default_mem = NULL;
}
+/*
+ * Return the offset into a iotlb slot required to keep the device happy.
+ */
+static unsigned int swiotlb_align_offset(struct device *dev, u64 addr)
+{
+ return addr & dma_get_min_align_mask(dev) & (IO_TLB_SIZE - 1);
+}
+
/*
* Bounce: copy the swiotlb buffer from or back to the original dma location
*/
@@ -346,10 +354,17 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size
size_t alloc_size = mem->slots[index].alloc_size;
unsigned long pfn = PFN_DOWN(orig_addr);
unsigned char *vaddr = phys_to_virt(tlb_addr);
+ unsigned int tlb_offset;
if (orig_addr == INVALID_PHYS_ADDR)
return;
+ tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) -
+ swiotlb_align_offset(dev, orig_addr);
+
+ orig_addr += tlb_offset;
+ alloc_size -= tlb_offset;
+
if (size > alloc_size) {
dev_WARN_ONCE(dev, 1,
"Buffer overflow detected. Allocation size: %zu. Mapping size: %zu.\n",
@@ -390,14 +405,6 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size
#define slot_addr(start, idx) ((start) + ((idx) << IO_TLB_SHIFT))
-/*
- * Return the offset into a iotlb slot required to keep the device happy.
- */
-static unsigned int swiotlb_align_offset(struct device *dev, u64 addr)
-{
- return addr & dma_get_min_align_mask(dev) & (IO_TLB_SIZE - 1);
-}
-
/*
* Carefully handle integer overflow which can occur when boundary_mask == ~0UL.
*/
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
[not found] ` <CGME20210628065823epcas2p19305f8b888a7fc0e883ec51db61e3bae@epcas2p1.samsung.com>
@ 2021-06-28 6:59 ` Chanho Park
2021-06-28 9:05 ` Greg KH
0 siblings, 1 reply; 15+ messages in thread
From: Chanho Park @ 2021-06-28 6:59 UTC (permalink / raw)
To: gregkh
Cc: Bumyong Lee, Chanho Park, Christoph Hellwig, Dominique MARTINET,
Horia Geantă,
stable, Konrad Rzeszutek Wilk
From: Bumyong Lee <bumyong.lee@samsung.com>
commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream.
(Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/20210301074436.919889-1-hch@lst.de/)
in case of driver wants to sync part of ranges with offset,
swiotlb_tbl_sync_single() copies from orig_addr base to tlb_addr with
offset and ends up with data mismatch.
It was removed from
"swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single",
but said logic has to be added back in.
From Linus's email:
"That commit which the removed the offset calculation entirely, because the old
(unsigned long)tlb_addr & (IO_TLB_SIZE - 1)
was wrong, but instead of removing it, I think it should have just
fixed it to be
(tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
instead. That way the slot offset always matches the slot index calculation."
(Unfortunatly that broke NVMe).
The use-case that drivers are hitting is as follow:
1. Get dma_addr_t from dma_map_single()
dma_addr_t tlb_addr = dma_map_single(dev, vaddr, vsize, DMA_TO_DEVICE);
|<---------------vsize------------->|
+-----------------------------------+
| | original buffer
+-----------------------------------+
vaddr
swiotlb_align_offset
|<----->|<---------------vsize------------->|
+-------+-----------------------------------+
| | | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
2. Do something
3. Sync dma_addr_t through dma_sync_single_for_device(..)
dma_sync_single_for_device(dev, tlb_addr + offset, size, DMA_TO_DEVICE);
Error case.
Copy data to original buffer but it is from base addr (instead of
base addr + offset) in original buffer:
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- size ->|
+-----------------------------------+
|##########| | original buffer
+-----------------------------------+
vaddr
The fix is to copy the data to the original buffer and take into
account the offset, like so:
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- offset ->|<- size ->|
+-----------------------------------+
| |##########| | original buffer
+-----------------------------------+
vaddr
[One fix which was Linus's that made more sense to as it created a
symmetry would break NVMe. The reason for that is the:
unsigned int offset = (tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
would come up with the proper offset, but it would lose the
alignment (which this patch contains).]
Fixes: 16fc3cef33a0 ("swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single")
Signed-off-by: Bumyong Lee <bumyong.lee@samsung.com>
Signed-off-by: Chanho Park <chanho61.park@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: Dominique MARTINET <dominique.martinet@atmark-techno.com>
Reported-by: Horia Geantă <horia.geanta@nxp.com>
Tested-by: Horia Geantă <horia.geanta@nxp.com>
CC: stable@vger.kernel.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
kernel/dma/swiotlb.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index fe4c01c14ab2..e96f3808e431 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -724,11 +724,17 @@ void swiotlb_tbl_sync_single(struct device *hwdev, phys_addr_t tlb_addr,
int index = (tlb_addr - io_tlb_start) >> IO_TLB_SHIFT;
size_t orig_size = io_tlb_orig_size[index];
phys_addr_t orig_addr = io_tlb_orig_addr[index];
+ unsigned int tlb_offset;
if (orig_addr == INVALID_PHYS_ADDR)
return;
- validate_sync_size_and_truncate(hwdev, orig_size, &size);
+ tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) -
+ swiotlb_align_offset(hwdev, orig_addr);
+
+ orig_addr += tlb_offset;
+
+ validate_sync_size_and_truncate(hwdev, orig_size - tlb_offset, &size);
switch (target) {
case SYNC_FOR_CPU:
--
2.32.0
^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-06-28 6:59 ` [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset Chanho Park
@ 2021-06-28 9:05 ` Greg KH
2021-06-28 9:25 ` Chanho Park
0 siblings, 1 reply; 15+ messages in thread
From: Greg KH @ 2021-06-28 9:05 UTC (permalink / raw)
To: Chanho Park
Cc: Bumyong Lee, Christoph Hellwig, Dominique MARTINET,
Horia Geantă,
stable, Konrad Rzeszutek Wilk
On Mon, Jun 28, 2021 at 03:59:16PM +0900, Chanho Park wrote:
> From: Bumyong Lee <bumyong.lee@samsung.com>
>
> commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream.
> (Backported as different form due to absence of below patch series
> https://lore.kernel.org/linux-iommu/20210301074436.919889-1-hch@lst.de/)
What stable kernel(s) is this for?
And did you send the same patch twice?
thanks,
greg k-h
^ permalink raw reply [flat|nested] 15+ messages in thread
* RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-06-28 9:05 ` Greg KH
@ 2021-06-28 9:25 ` Chanho Park
2021-06-28 9:36 ` 'Greg KH'
0 siblings, 1 reply; 15+ messages in thread
From: Chanho Park @ 2021-06-28 9:25 UTC (permalink / raw)
To: 'Greg KH'
Cc: 'Bumyong Lee', 'Christoph Hellwig',
'Dominique MARTINET', 'Horia Geantă',
stable, 'Konrad Rzeszutek Wilk'
> > commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream.
> > (Backported as different form due to absence of below patch series
> > https://lore.kernel.org/linux-iommu/20210301074436.919889-1-hch@lst.de/)
>
> What stable kernel(s) is this for?
Hmm. I sent this with "--in-reply-to" option of git send-email with your
e-mail's message-id but it didn't work well.
This is for linux-5.12.y tree.
>
> And did you send the same patch twice?
No. Unfortunately, they're different due to swiotlb patches. I backported
the patch to each kernel versions respectively.
Best Regards,
Chanho Park
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-06-28 9:25 ` Chanho Park
@ 2021-06-28 9:36 ` 'Greg KH'
2021-06-28 10:05 ` Chanho Park
2021-06-28 12:20 ` 'Greg KH'
0 siblings, 2 replies; 15+ messages in thread
From: 'Greg KH' @ 2021-06-28 9:36 UTC (permalink / raw)
To: Chanho Park
Cc: 'Bumyong Lee', 'Christoph Hellwig',
'Dominique MARTINET', 'Horia Geantă',
stable, 'Konrad Rzeszutek Wilk'
On Mon, Jun 28, 2021 at 06:25:28PM +0900, Chanho Park wrote:
> > > commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream.
> > > (Backported as different form due to absence of below patch series
> > > https://lore.kernel.org/linux-iommu/20210301074436.919889-1-hch@lst.de/)
> >
> > What stable kernel(s) is this for?
>
> Hmm. I sent this with "--in-reply-to" option of git send-email with your
> e-mail's message-id but it didn't work well.
That does not work when the message you are sending in-reply-to is no
longer in the recipient's message box :)
> This is for linux-5.12.y tree.
Great, thanks.
> > And did you send the same patch twice?
>
> No. Unfortunately, they're different due to swiotlb patches. I backported
> the patch to each kernel versions respectively.
What is the "other" patch for?
greg k-h
^ permalink raw reply [flat|nested] 15+ messages in thread
* RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-06-28 9:36 ` 'Greg KH'
@ 2021-06-28 10:05 ` Chanho Park
2021-06-28 12:20 ` 'Greg KH'
1 sibling, 0 replies; 15+ messages in thread
From: Chanho Park @ 2021-06-28 10:05 UTC (permalink / raw)
To: 'Greg KH'
Cc: 'Bumyong Lee', 'Christoph Hellwig',
'Dominique MARTINET', 'Horia Geantă',
stable, 'Konrad Rzeszutek Wilk'
> That does not work when the message you are sending in-reply-to is no
> longer in the recipient's message box :)
Ah. Got it.
>
> > This is for linux-5.12.y tree.
>
> Great, thanks.
>
> > > And did you send the same patch twice?
> >
> > No. Unfortunately, they're different due to swiotlb patches. I
> > backported the patch to each kernel versions respectively.
>
> What is the "other" patch for?
Another patch what I sent is for linux-5.10.y tree.
Regarding linux-5.10.y's patch and linux-5.12.y's patch, below patch was
applied only for linux-5.12.y tree and it made differences.
daf9514fd5eb swiotlb: Validate bounce size in the sync/unmap path
Best Regards,
Chanho Park
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-06-28 9:36 ` 'Greg KH'
2021-06-28 10:05 ` Chanho Park
@ 2021-06-28 12:20 ` 'Greg KH'
1 sibling, 0 replies; 15+ messages in thread
From: 'Greg KH' @ 2021-06-28 12:20 UTC (permalink / raw)
To: Chanho Park
Cc: 'Bumyong Lee', 'Christoph Hellwig',
'Dominique MARTINET', 'Horia Geantă',
stable, 'Konrad Rzeszutek Wilk'
On Mon, Jun 28, 2021 at 11:36:19AM +0200, 'Greg KH' wrote:
> On Mon, Jun 28, 2021 at 06:25:28PM +0900, Chanho Park wrote:
> > > > commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream.
> > > > (Backported as different form due to absence of below patch series
> > > > https://lore.kernel.org/linux-iommu/20210301074436.919889-1-hch@lst.de/)
> > >
> > > What stable kernel(s) is this for?
> >
> > Hmm. I sent this with "--in-reply-to" option of git send-email with your
> > e-mail's message-id but it didn't work well.
>
> That does not work when the message you are sending in-reply-to is no
> longer in the recipient's message box :)
>
> > This is for linux-5.12.y tree.
>
> Great, thanks.
>
> > > And did you send the same patch twice?
> >
> > No. Unfortunately, they're different due to swiotlb patches. I backported
> > the patch to each kernel versions respectively.
>
> What is the "other" patch for?
Looks like 5.10.y. Both now queued up, thanks.
greg k-h
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
[not found] ` <CGME20210628065652epcas2p28f39187d9c1519cea42bd99a7e9dd75d@epcas2p2.samsung.com>
@ 2021-06-28 6:57 ` Chanho Park
0 siblings, 0 replies; 15+ messages in thread
From: Chanho Park @ 2021-06-28 6:57 UTC (permalink / raw)
To: gregkh
Cc: Bumyong Lee, Chanho Park, Christoph Hellwig, Dominique MARTINET,
Horia Geantă,
stable, Konrad Rzeszutek Wilk
From: Bumyong Lee <bumyong.lee@samsung.com>
commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream.
(Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/20210301074436.919889-1-hch@lst.de/)
in case of driver wants to sync part of ranges with offset,
swiotlb_tbl_sync_single() copies from orig_addr base to tlb_addr with
offset and ends up with data mismatch.
It was removed from
"swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single",
but said logic has to be added back in.
From Linus's email:
"That commit which the removed the offset calculation entirely, because the old
(unsigned long)tlb_addr & (IO_TLB_SIZE - 1)
was wrong, but instead of removing it, I think it should have just
fixed it to be
(tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
instead. That way the slot offset always matches the slot index calculation."
(Unfortunatly that broke NVMe).
The use-case that drivers are hitting is as follow:
1. Get dma_addr_t from dma_map_single()
dma_addr_t tlb_addr = dma_map_single(dev, vaddr, vsize, DMA_TO_DEVICE);
|<---------------vsize------------->|
+-----------------------------------+
| | original buffer
+-----------------------------------+
vaddr
swiotlb_align_offset
|<----->|<---------------vsize------------->|
+-------+-----------------------------------+
| | | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
2. Do something
3. Sync dma_addr_t through dma_sync_single_for_device(..)
dma_sync_single_for_device(dev, tlb_addr + offset, size, DMA_TO_DEVICE);
Error case.
Copy data to original buffer but it is from base addr (instead of
base addr + offset) in original buffer:
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- size ->|
+-----------------------------------+
|##########| | original buffer
+-----------------------------------+
vaddr
The fix is to copy the data to the original buffer and take into
account the offset, like so:
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- offset ->|<- size ->|
+-----------------------------------+
| |##########| | original buffer
+-----------------------------------+
vaddr
[One fix which was Linus's that made more sense to as it created a
symmetry would break NVMe. The reason for that is the:
unsigned int offset = (tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
would come up with the proper offset, but it would lose the
alignment (which this patch contains).]
Fixes: 16fc3cef33a0 ("swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single")
Signed-off-by: Bumyong Lee <bumyong.lee@samsung.com>
Signed-off-by: Chanho Park <chanho61.park@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: Dominique MARTINET <dominique.martinet@atmark-techno.com>
Reported-by: Horia Geantă <horia.geanta@nxp.com>
Tested-by: Horia Geantă <horia.geanta@nxp.com>
CC: stable@vger.kernel.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
kernel/dma/swiotlb.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 0f61b14b0099..0ed0e1f215c7 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -667,6 +667,9 @@ void swiotlb_tbl_sync_single(struct device *hwdev, phys_addr_t tlb_addr,
if (orig_addr == INVALID_PHYS_ADDR)
return;
+ orig_addr += (tlb_addr & (IO_TLB_SIZE - 1)) -
+ swiotlb_align_offset(hwdev, orig_addr);
+
switch (target) {
case SYNC_FOR_CPU:
if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL))
--
2.32.0
^ permalink raw reply related [flat|nested] 15+ messages in thread
* RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-05-10 8:44 ` Christoph Hellwig
@ 2021-05-10 9:04 ` Chanho Park
-1 siblings, 0 replies; 15+ messages in thread
From: Chanho Park @ 2021-05-10 9:04 UTC (permalink / raw)
To: 'Christoph Hellwig'
Cc: 'Konrad Rzeszutek Wilk', 'Marek Szyprowski',
'Robin Murphy',
iommu, linux-kernel, 'Bumyong Lee'
(RESEND due to wrong encrypted message setting)
Hi,
> On Mon, May 10, 2021 at 05:30:57PM +0900, Chanho Park wrote:
> > +static unsigned int swiotlb_align_offset(struct device *dev, u64
> > +addr);
>
> Please just move swiotlb_align_offset up to avoid the forward declaration.
Okay. I'll move the position of the function next patch.
>
> > /*
> > * Bounce: copy the swiotlb buffer from or back to the original dma
> location
> > */
> > @@ -346,10 +347,17 @@ static void swiotlb_bounce(struct device *dev,
> phys_addr_t tlb_addr, size_t size
> > size_t alloc_size = mem->slots[index].alloc_size;
> > unsigned long pfn = PFN_DOWN(orig_addr);
> > unsigned char *vaddr = phys_to_virt(tlb_addr);
> > + unsigned int tlb_offset;
> >
> > if (orig_addr == INVALID_PHYS_ADDR)
> > return;
> >
> > + tlb_offset = (unsigned int)tlb_addr & (IO_TLB_SIZE - 1);
> > + tlb_offset -= swiotlb_align_offset(dev, orig_addr);
>
> Nit: I'd write this as:
>
> tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) -
> swiotlb_align_offset(dev, orig_addr);
>
> as there is no need for the cast, and just having a single assignment is
> easier to follow.
Great. It can be a single assignment as you suggested.
Best Regards,
Chanho Park
^ permalink raw reply [flat|nested] 15+ messages in thread
* RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
@ 2021-05-10 9:04 ` Chanho Park
0 siblings, 0 replies; 15+ messages in thread
From: Chanho Park @ 2021-05-10 9:04 UTC (permalink / raw)
To: 'Christoph Hellwig'
Cc: 'Konrad Rzeszutek Wilk',
linux-kernel, 'Bumyong Lee',
iommu, 'Robin Murphy'
(RESEND due to wrong encrypted message setting)
Hi,
> On Mon, May 10, 2021 at 05:30:57PM +0900, Chanho Park wrote:
> > +static unsigned int swiotlb_align_offset(struct device *dev, u64
> > +addr);
>
> Please just move swiotlb_align_offset up to avoid the forward declaration.
Okay. I'll move the position of the function next patch.
>
> > /*
> > * Bounce: copy the swiotlb buffer from or back to the original dma
> location
> > */
> > @@ -346,10 +347,17 @@ static void swiotlb_bounce(struct device *dev,
> phys_addr_t tlb_addr, size_t size
> > size_t alloc_size = mem->slots[index].alloc_size;
> > unsigned long pfn = PFN_DOWN(orig_addr);
> > unsigned char *vaddr = phys_to_virt(tlb_addr);
> > + unsigned int tlb_offset;
> >
> > if (orig_addr == INVALID_PHYS_ADDR)
> > return;
> >
> > + tlb_offset = (unsigned int)tlb_addr & (IO_TLB_SIZE - 1);
> > + tlb_offset -= swiotlb_align_offset(dev, orig_addr);
>
> Nit: I'd write this as:
>
> tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) -
> swiotlb_align_offset(dev, orig_addr);
>
> as there is no need for the cast, and just having a single assignment is
> easier to follow.
Great. It can be a single assignment as you suggested.
Best Regards,
Chanho Park
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
^ permalink raw reply [flat|nested] 15+ messages in thread
* RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
[not found] <CGME20210510085719epcas2p3b4467098ba64c4bdd51fe3ede38e3753@epcms2p4>
@ 2021-05-10 8:57 ` CHANHO PARK
0 siblings, 0 replies; 15+ messages in thread
From: CHANHO PARK @ 2021-05-10 8:57 UTC (permalink / raw)
[-- Attachment #1.1: Type: text/html, Size: 11295 bytes --]
[-- Attachment #2: body_20210510085719_1574384853.mht.pdf --]
[-- Type: application/octet-stream, Size: 13990 bytes --]
[-- Attachment #3: Type: text/plain, Size: 156 bytes --]
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
2021-05-10 8:30 ` Chanho Park
@ 2021-05-10 8:44 ` Christoph Hellwig
-1 siblings, 0 replies; 15+ messages in thread
From: Christoph Hellwig @ 2021-05-10 8:44 UTC (permalink / raw)
To: Chanho Park
Cc: Konrad Rzeszutek Wilk, Christoph Hellwig, Marek Szyprowski,
Robin Murphy, iommu, linux-kernel, Bumyong Lee
On Mon, May 10, 2021 at 05:30:57PM +0900, Chanho Park wrote:
> +static unsigned int swiotlb_align_offset(struct device *dev, u64 addr);
Please just move swiotlb_align_offset up to avoid the forward declaration.
> /*
> * Bounce: copy the swiotlb buffer from or back to the original dma location
> */
> @@ -346,10 +347,17 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size
> size_t alloc_size = mem->slots[index].alloc_size;
> unsigned long pfn = PFN_DOWN(orig_addr);
> unsigned char *vaddr = phys_to_virt(tlb_addr);
> + unsigned int tlb_offset;
>
> if (orig_addr == INVALID_PHYS_ADDR)
> return;
>
> + tlb_offset = (unsigned int)tlb_addr & (IO_TLB_SIZE - 1);
> + tlb_offset -= swiotlb_align_offset(dev, orig_addr);
Nit: I'd write this as:
tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) -
swiotlb_align_offset(dev, orig_addr);
as there is no need for the cast, and just having a single assignment
is easier to follow.
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
@ 2021-05-10 8:44 ` Christoph Hellwig
0 siblings, 0 replies; 15+ messages in thread
From: Christoph Hellwig @ 2021-05-10 8:44 UTC (permalink / raw)
To: Chanho Park
Cc: Konrad Rzeszutek Wilk, linux-kernel, Bumyong Lee, iommu,
Robin Murphy, Christoph Hellwig
On Mon, May 10, 2021 at 05:30:57PM +0900, Chanho Park wrote:
> +static unsigned int swiotlb_align_offset(struct device *dev, u64 addr);
Please just move swiotlb_align_offset up to avoid the forward declaration.
> /*
> * Bounce: copy the swiotlb buffer from or back to the original dma location
> */
> @@ -346,10 +347,17 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size
> size_t alloc_size = mem->slots[index].alloc_size;
> unsigned long pfn = PFN_DOWN(orig_addr);
> unsigned char *vaddr = phys_to_virt(tlb_addr);
> + unsigned int tlb_offset;
>
> if (orig_addr == INVALID_PHYS_ADDR)
> return;
>
> + tlb_offset = (unsigned int)tlb_addr & (IO_TLB_SIZE - 1);
> + tlb_offset -= swiotlb_align_offset(dev, orig_addr);
Nit: I'd write this as:
tlb_offset = (tlb_addr & (IO_TLB_SIZE - 1)) -
swiotlb_align_offset(dev, orig_addr);
as there is no need for the cast, and just having a single assignment
is easier to follow.
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
[not found] <CGME20210510083139epcas2p211d9bee16e5e8f8ea34e606c83ac3a55@epcas2p2.samsung.com>
@ 2021-05-10 8:30 ` Chanho Park
0 siblings, 0 replies; 15+ messages in thread
From: Chanho Park @ 2021-05-10 8:30 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Christoph Hellwig
Cc: Marek Szyprowski, Robin Murphy, iommu, linux-kernel, Bumyong Lee,
Chanho Park
From: Bumyong Lee <bumyong.lee@samsung.com>
in case of driver wants to sync part of ranges with offset,
swiotlb_tbl_sync_single() copies from orig_addr base to tlb_addr with offset
it makes data mismatch
it removed from "swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single",
but it have to be recovered
1. Get dma_addr_t from dma_map_single()
dma_addr_t tlb_addr = dma_map_single(dev, vaddr, vsize, DMA_TO_DEVICE);
|<---------------vsize------------->|
+-----------------------------------+
| | original buffer
+-----------------------------------+
vaddr
swiotlb_align_offset
|<----->|<---------------vsize------------->|
+-------+-----------------------------------+
| | | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
2. Do something
3. Sync dma_addr_t through dma_sync_single_for_device(..)
dma_sync_single_for_device(dev, tlb_addr + offset, size, DMA_TO_DEVICE);
Error case.
copy data to original buffer.
but it is from base addr in original buffer
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- size ->|
+-----------------------------------+
|##########| | original buffer
+-----------------------------------+
vaddr
FIX. copy data to original buffer.
but it is from base addr in original buffer
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- offset ->|<- size ->|
+-----------------------------------+
| |##########| | original buffer
+-----------------------------------+
vaddr
Fixes: 16fc3cef33a0 ("swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single")
Signed-off-by: Bumyong Lee <bumyong.lee@samsung.com>
Signed-off-by: Chanho Park <chanho61.park@samsung.com>
---
kernel/dma/swiotlb.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 8ca7d505d61c..e8243725e298 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -334,6 +334,7 @@ void __init swiotlb_exit(void)
io_tlb_default_mem = NULL;
}
+static unsigned int swiotlb_align_offset(struct device *dev, u64 addr);
/*
* Bounce: copy the swiotlb buffer from or back to the original dma location
*/
@@ -346,10 +347,17 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size
size_t alloc_size = mem->slots[index].alloc_size;
unsigned long pfn = PFN_DOWN(orig_addr);
unsigned char *vaddr = phys_to_virt(tlb_addr);
+ unsigned int tlb_offset;
if (orig_addr == INVALID_PHYS_ADDR)
return;
+ tlb_offset = (unsigned int)tlb_addr & (IO_TLB_SIZE - 1);
+ tlb_offset -= swiotlb_align_offset(dev, orig_addr);
+
+ orig_addr += tlb_offset;
+ alloc_size -= tlb_offset;
+
if (size > alloc_size) {
dev_WARN_ONCE(dev, 1,
"Buffer overflow detected. Allocation size: %zu. Mapping size: %zu.\n",
--
2.31.1
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset
@ 2021-05-10 8:30 ` Chanho Park
0 siblings, 0 replies; 15+ messages in thread
From: Chanho Park @ 2021-05-10 8:30 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Christoph Hellwig
Cc: linux-kernel, Bumyong Lee, iommu, Chanho Park, Robin Murphy
From: Bumyong Lee <bumyong.lee@samsung.com>
in case of driver wants to sync part of ranges with offset,
swiotlb_tbl_sync_single() copies from orig_addr base to tlb_addr with offset
it makes data mismatch
it removed from "swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single",
but it have to be recovered
1. Get dma_addr_t from dma_map_single()
dma_addr_t tlb_addr = dma_map_single(dev, vaddr, vsize, DMA_TO_DEVICE);
|<---------------vsize------------->|
+-----------------------------------+
| | original buffer
+-----------------------------------+
vaddr
swiotlb_align_offset
|<----->|<---------------vsize------------->|
+-------+-----------------------------------+
| | | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
2. Do something
3. Sync dma_addr_t through dma_sync_single_for_device(..)
dma_sync_single_for_device(dev, tlb_addr + offset, size, DMA_TO_DEVICE);
Error case.
copy data to original buffer.
but it is from base addr in original buffer
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- size ->|
+-----------------------------------+
|##########| | original buffer
+-----------------------------------+
vaddr
FIX. copy data to original buffer.
but it is from base addr in original buffer
swiotlb_align_offset
|<----->|<- offset ->|<- size ->|
+-------+-----------------------------------+
| | |##########| | swiotlb buffer
+-------+-----------------------------------+
tlb_addr
|<- offset ->|<- size ->|
+-----------------------------------+
| |##########| | original buffer
+-----------------------------------+
vaddr
Fixes: 16fc3cef33a0 ("swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single")
Signed-off-by: Bumyong Lee <bumyong.lee@samsung.com>
Signed-off-by: Chanho Park <chanho61.park@samsung.com>
---
kernel/dma/swiotlb.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 8ca7d505d61c..e8243725e298 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -334,6 +334,7 @@ void __init swiotlb_exit(void)
io_tlb_default_mem = NULL;
}
+static unsigned int swiotlb_align_offset(struct device *dev, u64 addr);
/*
* Bounce: copy the swiotlb buffer from or back to the original dma location
*/
@@ -346,10 +347,17 @@ static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size
size_t alloc_size = mem->slots[index].alloc_size;
unsigned long pfn = PFN_DOWN(orig_addr);
unsigned char *vaddr = phys_to_virt(tlb_addr);
+ unsigned int tlb_offset;
if (orig_addr == INVALID_PHYS_ADDR)
return;
+ tlb_offset = (unsigned int)tlb_addr & (IO_TLB_SIZE - 1);
+ tlb_offset -= swiotlb_align_offset(dev, orig_addr);
+
+ orig_addr += tlb_offset;
+ alloc_size -= tlb_offset;
+
if (size > alloc_size) {
dev_WARN_ONCE(dev, 1,
"Buffer overflow detected. Allocation size: %zu. Mapping size: %zu.\n",
--
2.31.1
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
^ permalink raw reply related [flat|nested] 15+ messages in thread
end of thread, other threads:[~2021-06-28 21:48 UTC | newest]
Thread overview: 15+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-25 9:26 FAILED: patch "[PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset" failed to apply to 5.12-stable tree gregkh
[not found] ` <CGME20210628065823epcas2p19305f8b888a7fc0e883ec51db61e3bae@epcas2p1.samsung.com>
2021-06-28 6:59 ` [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset Chanho Park
2021-06-28 9:05 ` Greg KH
2021-06-28 9:25 ` Chanho Park
2021-06-28 9:36 ` 'Greg KH'
2021-06-28 10:05 ` Chanho Park
2021-06-28 12:20 ` 'Greg KH'
-- strict thread matches above, loose matches on Subject: below --
2021-06-25 9:26 FAILED: patch "[PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset" failed to apply to 5.10-stable tree gregkh
[not found] ` <CGME20210628065652epcas2p28f39187d9c1519cea42bd99a7e9dd75d@epcas2p2.samsung.com>
2021-06-28 6:57 ` [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset Chanho Park
[not found] <CGME20210510085719epcas2p3b4467098ba64c4bdd51fe3ede38e3753@epcms2p4>
2021-05-10 8:57 ` CHANHO PARK
[not found] <CGME20210510083139epcas2p211d9bee16e5e8f8ea34e606c83ac3a55@epcas2p2.samsung.com>
2021-05-10 8:30 ` Chanho Park
2021-05-10 8:30 ` Chanho Park
2021-05-10 8:44 ` Christoph Hellwig
2021-05-10 8:44 ` Christoph Hellwig
2021-05-10 9:04 ` Chanho Park
2021-05-10 9:04 ` Chanho Park
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.