2023-04-13 15:41:52

by Michael Kelley (LINUX)

[permalink] [raw]
Subject: [PATCH 1/1] swiotlb: Fix debugfs reporting of reserved memory pools

For io_tlb_nslabs, the debugfs code reports the correct value for a
specific reserved memory pool. But for io_tlb_used, the value reported
is always for the default pool, not the specific reserved pool. Fix this.

Fixes: 5c850d31880e ("swiotlb: fix passing local variable to debugfs_create_ulong()")
Signed-off-by: Michael Kelley <[email protected]>
---

I don't have a way to test this directly with OF reserved memory, but I
cobbled together a hack to call rmem_swiotlb_device_init() multiple times
for different size reserved pools. I verified that reserved pool debugfs entries
are created as expected and that the value of io_tlb_used is *not* the value
from the default pool.

kernel/dma/swiotlb.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index dac42a2..db43de82 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -930,7 +930,9 @@ bool is_swiotlb_active(struct device *dev)

static int io_tlb_used_get(void *data, u64 *val)
{
- *val = mem_used(&io_tlb_default_mem);
+ struct io_tlb_mem *mem = data;
+
+ *val = mem_used(mem);
return 0;
}
DEFINE_DEBUGFS_ATTRIBUTE(fops_io_tlb_used, io_tlb_used_get, NULL, "%llu\n");
@@ -943,7 +945,7 @@ static void swiotlb_create_debugfs_files(struct io_tlb_mem *mem,
return;

debugfs_create_ulong("io_tlb_nslabs", 0400, mem->debugfs, &mem->nslabs);
- debugfs_create_file("io_tlb_used", 0400, mem->debugfs, NULL,
+ debugfs_create_file("io_tlb_used", 0400, mem->debugfs, mem,
&fops_io_tlb_used);
}

--
1.8.3.1


2023-04-13 19:01:52

by Petr Tesařík

[permalink] [raw]
Subject: Re: [PATCH 1/1] swiotlb: Fix debugfs reporting of reserved memory pools

On Thu, 13 Apr 2023 08:37:30 -0700
Michael Kelley <[email protected]> wrote:

> For io_tlb_nslabs, the debugfs code reports the correct value for a
> specific reserved memory pool. But for io_tlb_used, the value reported
> is always for the default pool, not the specific reserved pool. Fix this.
>
> Fixes: 5c850d31880e ("swiotlb: fix passing local variable to debugfs_create_ulong()")
> Signed-off-by: Michael Kelley <[email protected]>
> ---
>
> I don't have a way to test this directly with OF reserved memory, but I
> cobbled together a hack to call rmem_swiotlb_device_init() multiple times
> for different size reserved pools. I verified that reserved pool debugfs entries
> are created as expected and that the value of io_tlb_used is *not* the value
> from the default pool.

Yeah, the only in-tree user is Mediatek Asurada, and I don't have one
either...

The patch looks good to me. But you know, I'm now well-known for
breaking things because of insufficient testing. ;-)

Petr T

> kernel/dma/swiotlb.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
> index dac42a2..db43de82 100644
> --- a/kernel/dma/swiotlb.c
> +++ b/kernel/dma/swiotlb.c
> @@ -930,7 +930,9 @@ bool is_swiotlb_active(struct device *dev)
>
> static int io_tlb_used_get(void *data, u64 *val)
> {
> - *val = mem_used(&io_tlb_default_mem);
> + struct io_tlb_mem *mem = data;
> +
> + *val = mem_used(mem);
> return 0;
> }
> DEFINE_DEBUGFS_ATTRIBUTE(fops_io_tlb_used, io_tlb_used_get, NULL,
> "%llu\n"); @@ -943,7 +945,7 @@ static void
> swiotlb_create_debugfs_files(struct io_tlb_mem *mem, return;
>
> debugfs_create_ulong("io_tlb_nslabs", 0400, mem->debugfs,
> &mem->nslabs);
> - debugfs_create_file("io_tlb_used", 0400, mem->debugfs, NULL,
> + debugfs_create_file("io_tlb_used", 0400, mem->debugfs, mem,
> &fops_io_tlb_used);
> }
>

2023-04-16 06:28:55

by Christoph Hellwig

[permalink] [raw]
Subject: Re: [PATCH 1/1] swiotlb: Fix debugfs reporting of reserved memory pools

Thanks,

applied to the dma-mapping for-next tree.