FAILED: patch "[PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset" failed to apply to 5.12-stable tree

PATCH In review [email protected] 2 participants 7 messages Jun 25, 2021 – Jun 28, 2021

Subsystems (2)

Stable branch DMA mapping helpers

lore

Share & tools

lore

[email protected]

author maintainer

2021-06-25 09:26 UTC #0 lore raw

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 <[email protected]>.

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From 5f89468e2f060031cd89fd4287298e0eaf246bf6 Mon Sep 17 00:00:00 2001
From: Bumyong Lee <[email protected]>
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 <[email protected]>

Signed-off-by Chanho Park <[email protected]>

Reviewed-by Christoph Hellwig <[email protected]>

Reported-by Dominique MARTINET <[email protected]>

Reported-by Horia Geantă <[email protected]>

Tested-by Horia Geantă <[email protected]>

CC [email protected]

Signed-off-by Konrad Rzeszutek Wilk <[email protected]>

kernel/dma/swiotlb.c +15 -8

index 8ca7d505d61c..e50df8d8f87e 100644
@@ -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.
  */

Chanho Park

2021-06-28 08:58 UTC #1 lore raw

PATCH 1/1 swiotlb: manipulate orig_addr when tlb_addr has offset

From: Bumyong Lee <[email protected]>

commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream. (Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/[email protected]/)

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 <[email protected]>

Signed-off-by Chanho Park <[email protected]>

Reviewed-by Christoph Hellwig <[email protected]>

Reported-by Dominique MARTINET <[email protected]>

Reported-by Horia Geantă <[email protected]>

Tested-by Horia Geantă <[email protected]>

CC [email protected]

Signed-off-by Konrad Rzeszutek Wilk <[email protected]>

---
 kernel/dma/swiotlb.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

kernel/dma/swiotlb.c +7 -1

index fe4c01c14ab2..e96f3808e431 100644
@@ -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:

Signature · 1 line

2.32.0

Greg KH

author maintainer

2021-06-28 09:06 UTC #2 lore raw

Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset

Quoted 5 lines · Chanho Park

From: Bumyong Lee <[email protected]>
commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream. (Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/[email protected]/)

What stable kernel(s) is this for?

And did you send the same patch twice?

thanks,

greg k-h

Chanho Park

2021-06-28 09:30 UTC #3 lore raw

RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset

Quoted 5 lines

Quoted 3 lines
commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream. (Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/[email protected]/)
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.

Quoted 2 lines

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

'Greg KH'

author maintainer

2021-06-28 09:36 UTC #4 lore raw

Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset

Quoted 8 lines · Chanho Park

Quoted 5 lines
Quoted 3 lines
commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream. (Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/[email protected]/)
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 :)

Quoted 1 line

This is for linux-5.12.y tree.

Great, thanks.

Quoted 4 lines

Quoted 1 line
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

'Greg KH'

author maintainer

2021-06-28 12:20 UTC #5 lore raw

Re: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset

Quoted 15 lines · Greg KH

Quoted 8 lines · Chanho Park
Quoted 5 lines
Quoted 3 lines
commit 5f89468e2f060031cd89fd4287298e0eaf246bf6 upstream. (Backported as different form due to absence of below patch series
https://lore.kernel.org/linux-iommu/[email protected]/)
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 :)
Quoted 1 line
This is for linux-5.12.y tree.

Quoted 8 lines

Great, thanks.
Quoted 4 lines
Quoted 1 line
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

Chanho Park

2021-06-28 21:48 UTC #6 lore raw

RE: [PATCH] swiotlb: manipulate orig_addr when tlb_addr has offset

In reply to 'Greg KH'

Quoted 2 lines

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.

Quoted 11 lines

Quoted 2 lines
This is for linux-5.12.y tree.
Great, thanks.
Quoted 4 lines
Quoted 1 line
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