| Message ID | 1613501314-2392-1-git-send-email-jhugo@codeaurora.org (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
| Series | mhi_bus: core: Return EBUSY if MHI ring is full | expand |
On 2021-02-16 10:48 AM, Jeffrey Hugo wrote: > From: Fan Wu <wufan@codeaurora.org> > > Currently ENOMEM is returned when MHI ring is full. This error code is > very misleading. Change to EBUSY instead. > > Signed-off-by: Fan Wu <wufan@codeaurora.org> > Signed-off-by: Jeffrey Hugo <jhugo@codeaurora.org> > --- > drivers/bus/mhi/core/main.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c > index f182736..21eb5fc 100644 > --- a/drivers/bus/mhi/core/main.c > +++ b/drivers/bus/mhi/core/main.c > @@ -996,7 +996,7 @@ static int mhi_queue(struct mhi_device *mhi_dev, > struct mhi_buf_info *buf_info, > > ret = mhi_is_ring_full(mhi_cntrl, tre_ring); > if (unlikely(ret)) { > - ret = -ENOMEM; > + ret = -EBUSY; > goto exit_unlock; > } ENOMEM is descriptive of the state of the ring since you basically cannot queue any more packets as no memory is currently available. But I agree, it can be misleading for this API. How about EAGAIN in place of EBUSY, which tells the user to try the queue attempt again implying memory should become available as more elements are consumed by the device/client? Thanks, Bhaumik --- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On 2/16/2021 1:22 PM, Bhaumik Bhatt wrote: > On 2021-02-16 10:48 AM, Jeffrey Hugo wrote: >> From: Fan Wu <wufan@codeaurora.org> >> >> Currently ENOMEM is returned when MHI ring is full. This error code is >> very misleading. Change to EBUSY instead. >> >> Signed-off-by: Fan Wu <wufan@codeaurora.org> >> Signed-off-by: Jeffrey Hugo <jhugo@codeaurora.org> >> --- >> drivers/bus/mhi/core/main.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c >> index f182736..21eb5fc 100644 >> --- a/drivers/bus/mhi/core/main.c >> +++ b/drivers/bus/mhi/core/main.c >> @@ -996,7 +996,7 @@ static int mhi_queue(struct mhi_device *mhi_dev, >> struct mhi_buf_info *buf_info, >> >> ret = mhi_is_ring_full(mhi_cntrl, tre_ring); >> if (unlikely(ret)) { >> - ret = -ENOMEM; >> + ret = -EBUSY; >> goto exit_unlock; >> } > > ENOMEM is descriptive of the state of the ring since you basically > cannot queue any > more packets as no memory is currently available. > > But I agree, it can be misleading for this API. How about EAGAIN in > place of EBUSY, > which tells the user to try the queue attempt again implying memory > should become > available as more elements are consumed by the device/client? Fan and I think EAGAIN is fine. Will send a v2.
On Tue, Feb 16, 2021 at 11:48:34AM -0700, Jeffrey Hugo wrote: > From: Fan Wu <wufan@codeaurora.org> > > Currently ENOMEM is returned when MHI ring is full. This error code is > very misleading. Change to EBUSY instead. > Please use the subject prefix: "bus: mhi: ..." Thanks, Mani > Signed-off-by: Fan Wu <wufan@codeaurora.org> > Signed-off-by: Jeffrey Hugo <jhugo@codeaurora.org> > --- > drivers/bus/mhi/core/main.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c > index f182736..21eb5fc 100644 > --- a/drivers/bus/mhi/core/main.c > +++ b/drivers/bus/mhi/core/main.c > @@ -996,7 +996,7 @@ static int mhi_queue(struct mhi_device *mhi_dev, struct mhi_buf_info *buf_info, > > ret = mhi_is_ring_full(mhi_cntrl, tre_ring); > if (unlikely(ret)) { > - ret = -ENOMEM; > + ret = -EBUSY; > goto exit_unlock; > } > > -- > Qualcomm Technologies, Inc. is a member of the > Code Aurora Forum, a Linux Foundation Collaborative Project. >
On Tue, 16 Feb 2021 at 19:50, Jeffrey Hugo <jhugo@codeaurora.org> wrote: > > From: Fan Wu <wufan@codeaurora.org> > > Currently ENOMEM is returned when MHI ring is full. This error code is > very misleading. Change to EBUSY instead. Well, there is no space left in the ring, so it's no so misleading. Regards, Loic
On 2/17/2021 8:02 AM, Loic Poulain wrote: > On Tue, 16 Feb 2021 at 19:50, Jeffrey Hugo <jhugo@codeaurora.org> wrote: >> >> From: Fan Wu <wufan@codeaurora.org> >> >> Currently ENOMEM is returned when MHI ring is full. This error code is >> very misleading. Change to EBUSY instead. > > Well, there is no space left in the ring, so it's no so misleading. ENOMEM is typically a memory allocation failure which is not what a client is going to think of regarding the ring, and it's not a unique failure code in this case. gen_tre can also return ENOMEM, which makes it difficult for the client to know if there is some significant failure, or they might just need to wait (assuming that is something the client can do).
On Wed, 17 Feb 2021 at 16:06, Jeffrey Hugo <jhugo@codeaurora.org> wrote: > > On 2/17/2021 8:02 AM, Loic Poulain wrote: > > On Tue, 16 Feb 2021 at 19:50, Jeffrey Hugo <jhugo@codeaurora.org> wrote: > >> > >> From: Fan Wu <wufan@codeaurora.org> > >> > >> Currently ENOMEM is returned when MHI ring is full. This error code is > >> very misleading. Change to EBUSY instead. > > > > Well, there is no space left in the ring, so it's no so misleading. > > ENOMEM is typically a memory allocation failure which is not what a > client is going to think of regarding the ring, and it's not a unique > failure code in this case. gen_tre can also return ENOMEM, which makes > it difficult for the client to know if there is some significant > failure, or they might just need to wait (assuming that is something the > client can do). Yes, fair enough, I overlooked the other thread, -EAGAIN would indeed make sense. Regards, Loic
diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c index f182736..21eb5fc 100644 --- a/drivers/bus/mhi/core/main.c +++ b/drivers/bus/mhi/core/main.c @@ -996,7 +996,7 @@ static int mhi_queue(struct mhi_device *mhi_dev, struct mhi_buf_info *buf_info, ret = mhi_is_ring_full(mhi_cntrl, tre_ring); if (unlikely(ret)) { - ret = -ENOMEM; + ret = -EBUSY; goto exit_unlock; }