Re: [PATCH] PCI: epf-mhi: return 0 on success instead of positive jiffies

[Bjorn Helgaas <helgaas@kernel.org>]

On Fri, Feb 06, 2026 at 03:05:29PM -0500, Daniel Hodges wrote:
> wait_for_completion_timeout() returns the number of jiffies remaining
> on success (positive value) or 0 on timeout. The pci_epf_mhi_edma_read()
> and pci_epf_mhi_edma_write() functions use the return value directly as
> their own return value, only converting timeout (0) to -ETIMEDOUT.
> 
> On success, they return the positive jiffies value. The callers in
> drivers/bus/mhi/ep/ring.c check for errors with "if (ret < 0)" for
> read_sync and "if (ret)" for write_sync. This causes write_sync success
> cases to be incorrectly treated as errors since the positive jiffies
> value is non-zero.
> 
> Fix by setting ret to 0 when wait_for_completion_timeout() succeeds.
> 
> Fixes: 7b99aaaddabb ("PCI: epf-mhi: Add eDMA support")
> Cc: stable@vger.kernel.org
> Signed-off-by: Daniel Hodges <git@danielhodges.dev>

Thanks for the patch!

Two questions: first, is there any reason why __mhi_ep_cache_ring()
tests for "ret < 0" but mhi_ep_ring_add_element() tests for "ret"
(non-zero)?  Could/should they both test just for non-zero, which I
think is the typical style?

Second, the subject and commit log are perfectly correct but basically
at the level of describing the C code.  I propose something along
these lines:

  PCI: epf-mhi: Return 0, not remaining timeout, when eDMA ops complete

  pci_epf_mhi_edma_read() and pci_epf_mhi_edma_write() start DMA
  operations and wait for completion with a timeout.

  On successful completion, they previously returned the remaining
  timeout, which callers may treat as an error.  In particular,
  mhi_ep_ring_add_element(), which calls pci_epf_mhi_edma_write() via
  mhi_cntrl->write_sync(), interprets any non-zero return value as
  failure.

  Return 0 on success instead of the remaining timeout to prevent
  mhi_ep_ring_add_element() from treating successful completion as an
  error.

> ---
>  drivers/pci/endpoint/functions/pci-epf-mhi.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/pci/endpoint/functions/pci-epf-mhi.c b/drivers/pci/endpoint/functions/pci-epf-mhi.c
> index 6643a88c7a0c..2f077d0b7957 100644
> --- a/drivers/pci/endpoint/functions/pci-epf-mhi.c
> +++ b/drivers/pci/endpoint/functions/pci-epf-mhi.c
> @@ -367,6 +367,8 @@ static int pci_epf_mhi_edma_read(struct mhi_ep_cntrl *mhi_cntrl,
>  		dev_err(dev, "DMA transfer timeout\n");
>  		dmaengine_terminate_sync(chan);
>  		ret = -ETIMEDOUT;
> +	} else {
> +		ret = 0;
>  	}
>  
>  err_unmap:
> @@ -438,6 +440,8 @@ static int pci_epf_mhi_edma_write(struct mhi_ep_cntrl *mhi_cntrl,
>  		dev_err(dev, "DMA transfer timeout\n");
>  		dmaengine_terminate_sync(chan);
>  		ret = -ETIMEDOUT;
> +	} else {
> +		ret = 0;
>  	}
>  
>  err_unmap:
> -- 
> 2.52.0
>