[PATCH] cxl: Don't ignore add_process_element result when attaching context

Previously, when attaching a context in dedicated mode, we ignored
the result of add_process_element, which could potentially fail.

If add_process_element returns and error, pass it back to the caller.

Signed-off-by: Daniel Axtens <redacted>

---

I'm not sure if this is worth sending to stable: it doesn't fix
any actual bug, it just lets us know earlier if we've failed.

However, if mpe, Mikey or Ian feel like it should go to stable I
have no objection to that either.
---
 drivers/misc/cxl/native.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/misc/cxl/native.c b/drivers/misc/cxl/native.c
index debd97147b58..22c06d91faa8 100644
--- a/drivers/misc/cxl/native.c
+++ b/drivers/misc/cxl/native.c

@@ -551,9 +551,7 @@ static int attach_afu_directed(struct cxl_context *ctx, u64 wed, u64 amr)
 	if ((result = cxl_afu_check_and_enable(ctx->afu)))
 		return result;
 
-	add_process_element(ctx);
-
-	return 0;
+	return add_process_element(ctx);
 }
 
 static int deactivate_afu_directed(struct cxl_afu *afu)

-- 
2.1.4

On Wed, 2015-07-29 at 14:07 +1000, Daniel Axtens wrote:

Previously, when attaching a context in dedicated mode, we ignored

Previously? You mean currently? Now is before this patch is applied.

the result of add_process_element, which could potentially fail.

If add_process_element returns and error, pass it back to the caller.

                                   ^

I can fix.

Signed-off-by: Daniel Axtens <redacted>

---

I'm not sure if this is worth sending to stable: it doesn't fix
any actual bug, it just lets us know earlier if we've failed.

However, if mpe, Mikey or Ian feel like it should go to stable I
have no objection to that either.

stable_kernel_rules.txt says:

 - It must fix a real bug that bothers people (not a, "This could be a
   problem..." type thing).


So I think not. But rules are flexible so I can be convinced otherwise if
anyone has a good argument.

cheers

On Wed, 2015-07-29 at 14:25 +1000, Michael Ellerman wrote:

On Wed, 2015-07-29 at 14:07 +1000, Daniel Axtens wrote:

quoted

Previously, when attaching a context in dedicated mode, we ignored

Previously? You mean currently? Now is before this patch is applied.

I do mean currently. Apologies for the temporal confusion :)

quoted

the result of add_process_element, which could potentially fail.

If add_process_element returns and error, pass it back to the caller.

                                   ^

I can fix.

Many thanks: it's good to know that my spelling mistakes will only be
immortalised in the list archives rather than in the kernel's git
history. :p


-- 
Regards,
Daniel

On Wed, 2015-07-29 at 14:40 +1000, Daniel Axtens wrote:

On Wed, 2015-07-29 at 14:25 +1000, Michael Ellerman wrote:

quoted

On Wed, 2015-07-29 at 14:07 +1000, Daniel Axtens wrote:

quoted

Previously, when attaching a context in dedicated mode, we ignored

=20
Previously? You mean currently? Now is before this patch is applied.

I do mean currently. Apologies for the temporal confusion :)
=20

quoted

=20

quoted

the result of add_process_element, which could potentially fail.
=20
If add_process_element returns and error, pass it back to the caller.

                                   ^
=20
I can fix.
=20

Many thanks: it's good to know that my spelling mistakes will only be
immortalised in the list archives rather than in the kernel's git
history. :p

Not to mention the double negative in the patch subject! "Don't
ignore"... come on!?!?  :-P

Mikey

On Wed, 2015-29-07 at 04:07:22 UTC, Daniel Axtens wrote:

Previously, when attaching a context in dedicated mode, we ignored
the result of add_process_element, which could potentially fail.

If add_process_element returns and error, pass it back to the caller.

Signed-off-by: Daniel Axtens <redacted>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/368857c16c595eb7537c

cheers