[PATCH] ixgbe: Don't call kmap() on page allocated with GFP_ATOMIC

[PATCH] ixgbe: Don't call kmap() on page allocated with GFP_ATOMIC

[Fabio M. De Francesco]

Pages allocated with GFP_ATOMIC cannot come from Highmem. This is why
there is no need to call kmap() on them.

Therefore, don't call kmap() on rx_buffer->page() and instead use a
plain page_address() to get the kernel address.

Suggested-by: Ira Weiny <ira.weiny@intel.com>
Suggested-by: Alexander Duyck <alexander.duyck@gmail.com>
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index 628d0eb0599f..71196fd92f81 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -1966,15 +1966,13 @@ static bool ixgbe_check_lbtest_frame(struct ixgbe_rx_buffer *rx_buffer,
 
 	frame_size >>= 1;
 
-	data = kmap(rx_buffer->page) + rx_buffer->page_offset;
+	data = page_address(rx_buffer->page) + rx_buffer->page_offset;
 
 	if (data[3] != 0xFF ||
 	    data[frame_size + 10] != 0xBE ||
 	    data[frame_size + 12] != 0xAF)
 		match = false;
 
-	kunmap(rx_buffer->page);
-
 	return match;
 }
 
-- 
2.36.1

Re: [PATCH] ixgbe: Don't call kmap() on page allocated with GFP_ATOMIC

[Ira Weiny]

On Mon, Jul 04, 2022 at 04:01:29PM +0200, Fabio M. De Francesco wrote:
> Pages allocated with GFP_ATOMIC cannot come from Highmem. This is why
> there is no need to call kmap() on them.

I'm still not 100% sure where this page gets allocated but AFAICT it is
allocated in ixgbe_alloc_mapped_page() which calls dev_alloc_pages() for the
allocation which is where the GFP_ATOMIC is specified.

I think I would add this detail here.

That said, and assuming my analysis is correct, the code looks fine so:

Reviewed-by: Ira Weiny <ira.weiny@intel.com>

> 
> Therefore, don't call kmap() on rx_buffer->page() and instead use a
> plain page_address() to get the kernel address.
> 
> Suggested-by: Ira Weiny <ira.weiny@intel.com>
> Suggested-by: Alexander Duyck <alexander.duyck@gmail.com>
> Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
> ---
>  drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
> index 628d0eb0599f..71196fd92f81 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
> @@ -1966,15 +1966,13 @@ static bool ixgbe_check_lbtest_frame(struct ixgbe_rx_buffer *rx_buffer,
>  
>  	frame_size >>= 1;
>  
> -	data = kmap(rx_buffer->page) + rx_buffer->page_offset;
> +	data = page_address(rx_buffer->page) + rx_buffer->page_offset;
>  
>  	if (data[3] != 0xFF ||
>  	    data[frame_size + 10] != 0xBE ||
>  	    data[frame_size + 12] != 0xAF)
>  		match = false;
>  
> -	kunmap(rx_buffer->page);
> -
>  	return match;
>  }
>  
> -- 
> 2.36.1
> 

Re: [PATCH] ixgbe: Don't call kmap() on page allocated with GFP_ATOMIC

[Alexander Duyck]

On Tue, Jul 5, 2022 at 11:22 AM Ira Weiny <ira.weiny@intel.com> wrote:
>
> On Mon, Jul 04, 2022 at 04:01:29PM +0200, Fabio M. De Francesco wrote:
> > Pages allocated with GFP_ATOMIC cannot come from Highmem. This is why
> > there is no need to call kmap() on them.
>
> I'm still not 100% sure where this page gets allocated but AFAICT it is
> allocated in ixgbe_alloc_mapped_page() which calls dev_alloc_pages() for the
> allocation which is where the GFP_ATOMIC is specified.
>
> I think I would add this detail here.
>
> That said, and assuming my analysis is correct, the code looks fine so:

Yeah, this is actually called out in other spots in the buffer
cleaning path. This is just something I had overlooked and left in
place back a few refactors ago.. :-)

https://elixir.bootlin.com/linux/latest/source/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c#L1795

Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>

RE: [Intel-wired-lan] [PATCH] ixgbe: Don't call kmap() on page allocated with GFP_ATOMIC

[G, GurucharanX]

> -----Original Message-----
> From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf Of
> Fabio M. De Francesco
> Sent: Monday, July 4, 2022 7:31 PM
> To: Brandeburg, Jesse <jesse.brandeburg@intel.com>; Nguyen, Anthony L
> <anthony.l.nguyen@intel.com>; David S. Miller <davem@davemloft.net>;
> Eric Dumazet <edumazet@google.com>; Jakub Kicinski <kuba@kernel.org>;
> Paolo Abeni <pabeni@redhat.com>; Alexei Starovoitov <ast@kernel.org>;
> Daniel Borkmann <daniel@iogearbox.net>; Jesper Dangaard Brouer
> <hawk@kernel.org>; John Fastabend <john.fastabend@gmail.com>; intel-
> wired-lan@lists.osuosl.org; netdev@vger.kernel.org; linux-
> kernel@vger.kernel.org; bpf@vger.kernel.org
> Cc: Weiny, Ira <ira.weiny@intel.com>; Fabio M. De Francesco
> <fmdefrancesco@gmail.com>
> Subject: [Intel-wired-lan] [PATCH] ixgbe: Don't call kmap() on page allocated
> with GFP_ATOMIC
> 
> Pages allocated with GFP_ATOMIC cannot come from Highmem. This is why
> there is no need to call kmap() on them.
> 
> Therefore, don't call kmap() on rx_buffer->page() and instead use a plain
> page_address() to get the kernel address.
> 
> Suggested-by: Ira Weiny <ira.weiny@intel.com>
> Suggested-by: Alexander Duyck <alexander.duyck@gmail.com>
> Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
> ---
>  drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 

Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)