driver: crypto: hisilicon: fix error handling

[yuzexiyzx]

1. support num = 0 in qm_set_vt_common().
2. release qp when qm_xqc_cfg() failed.
3. add check of msg and qp->num in hisi_qp_recv_sync().

[jenswi-linaro]

There's a typo in the title of the commit, I guess it should be "exception".

Please explain in plain text in the commit message why you're doing this. Most of what you're doing can be seen in the diff, but sometimes it's nice with a summary of that too.

[yuzexiyzx]

@jenswi-linaro comment solved.
changed to

1. support num = 0 in qm_set_vft_common().
   When qm_set_xqc_set() failed, it will set q_num to zero.
   qm_set_vft_common() need to support this scene.
2. release qp when qm_xqc_cfg() failed.
3. add check of msg and qp->num in hisi_qp_recv_sync().
   hisi_qp_recv_sync() is not static function, we need to
   check every input parameter to ensure that it is safe.

[jenswi-linaro]

I'm sorry I think you missed my point. You're in principle describing the diff in detail. In a good commit message you describe why a change is being made, not just what's changing.

If you need to make a list of all things done in a patch it's usually a sign that it should have been multiple patches, but in this case, it seems that it all belongs in one patch.

How about something like:

driver: crypto: hisilicon: fix error handling

When the XXX fails to configure, qm_set_xqc_vft() is called with the num argument
as zero to disable the device (or whatever happens). Update qm_set_xqc_vft() to
handle this error path.

[etienne-lms]

since it's an error path, prefer have err_ prefix in the branch label.

[yuzexiyzx]

solved

[yuzexiyzx]

@jenswi-linaro @etienne-lms All comments have been solved.

[etienne-lms]

Acked-by: Etienne Carriere <etienne.carriere@foss.st.com>
It think the IBART test failure is unrelated to this change.

[jenswi-linaro]

The title of the commit says:

driver: crypto: hisilicon: fix exception handling

I think the word "exception" might be misleading since we're not dealing with what we normally call exceptions in this patch. How about using the word "error" instead?

[jenswi-linaro]

With my comment addressed, please apply:
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

[yuzexiyzx]

@jforissier All comments have been solved.

[jforissier]

The IBART error can be ignored (rebase needed). @yuzexiyzx thanks!