[RHEL6 PATCH] iwlwifi: remove spurious warning

[RHEL6 PATCH] iwlwifi: remove spurious warning

[Stanislaw Gruszka]

BZ#837754

Description:
Some devices do not set SEQ_RX_FRAME flags for few commands that should
not be reclaimed. It was supposed to be fixed in firmware, but it was
not.

Upstream change also add possibility to configure which commands should
not be reclaimed, but that option is not used - only original 6 commands
are configured. So I did not backport that part, because it make things
somewhat (not necessarily) complex.

Upstream:
commit d663ee73f62903aed334eca25832a76c4a7e4c1c
Author: Johannes Berg <johannes.berg@intel.com>
Date:   Sat Mar 10 13:00:07 2012 -0800

    iwlwifi: abstract out missing SEQ_RX_FRAME workaround

Testing:
Compiled only.
---
 drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c |   11 -----------
 1 files changed, 0 insertions(+), 11 deletions(-)

diff --git a/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c b/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c
index 1daf01e..80e9328 100644
--- a/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c
+++ b/drivers/net/wireless/iwlwifi/iwl-trans-pcie-rx.c
@@ -449,17 +449,6 @@ static void iwl_rx_handle(struct iwl_trans *trans)
 		else
 			cmd = NULL;
 
-		/* warn if this is cmd response / notification and the uCode
-		 * didn't set the SEQ_RX_FRAME for a frame that is
-		 * uCode-originated
-		 * If you saw this code after the second half of 2012, then
-		 * please remove it
-		 */
-		WARN(pkt->hdr.cmd != REPLY_TX && reclaim == false &&
-		     (!(pkt->hdr.sequence & SEQ_RX_FRAME)),
-		     "reclaim is false, SEQ_RX_FRAME unset: %s\n",
-		     get_cmd_string(pkt->hdr.cmd));
-
 		err = iwl_rx_dispatch(priv(trans), rxb, cmd);
 
 		/*
-- 
1.7.1

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Johannes Berg]

On Mon, 2012-07-09 at 13:48 +0200, Stanislaw Gruszka wrote:
> BZ#837754
> 
> Description:
> Some devices do not set SEQ_RX_FRAME flags for few commands that should
> not be reclaimed. It was supposed to be fixed in firmware, but it was
> not.
> 
> Upstream change also add possibility to configure which commands should
> not be reclaimed, but that option is not used - only original 6 commands
> are configured. So I did not backport that part, because it make things
> somewhat (not necessarily) complex.

Yeah, looks fine. I'd hoped it was fixed for iwlwifi drivers (we
couldn't remember if it was 4965 only or not) but it turns out it wasn't
so this is needed.

I don't know if there's anything I should do about this though?

johannes

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Stanislaw Gruszka]

On Mon, Jul 09, 2012 at 01:50:47PM +0200, Johannes Berg wrote:
> On Mon, 2012-07-09 at 13:48 +0200, Stanislaw Gruszka wrote:
> > BZ#837754
> > 
> > Description:
> > Some devices do not set SEQ_RX_FRAME flags for few commands that should
> > not be reclaimed. It was supposed to be fixed in firmware, but it was
> > not.
> > 
> > Upstream change also add possibility to configure which commands should
> > not be reclaimed, but that option is not used - only original 6 commands
> > are configured. So I did not backport that part, because it make things
> > somewhat (not necessarily) complex.
> 
> Yeah, looks fine. I'd hoped it was fixed for iwlwifi drivers (we
> couldn't remember if it was 4965 only or not) but it turns out it wasn't
> so this is needed.
> 
> I don't know if there's anything I should do about this though?

Sorry Johannes that was not intended to go to you and to the public.

Stanislaw

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Emmanuel Grumbach]

>> > BZ#837754
>> >
>> > Description:
>> > Some devices do not set SEQ_RX_FRAME flags for few commands that should
>> > not be reclaimed. It was supposed to be fixed in firmware, but it was
>> > not.
>> >
>> > Upstream change also add possibility to configure which commands should
>> > not be reclaimed, but that option is not used - only original 6 commands
>> > are configured. So I did not backport that part, because it make things
>> > somewhat (not necessarily) complex.
>>
>> Yeah, looks fine. I'd hoped it was fixed for iwlwifi drivers (we
>> couldn't remember if it was 4965 only or not) but it turns out it wasn't
>> so this is needed.
>>
>> I don't know if there's anything I should do about this though?
>
> Sorry Johannes that was not intended to go to you and to the public.
>
> Stanislaw
> --

I would be glad to know on what commands this WARN_ON triggers
though... The command name should be printed along with the WARN_ON.
Also it is not really a bug in fw, it is more like a convention that
we (Linux driver developers) don't really like because it creates a
dependency between 2 layers in the driver, but this is another story.

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Stanislaw Gruszka]

On Mon, Jul 09, 2012 at 10:10:38PM +0300, Emmanuel Grumbach wrote:
> I would be glad to know on what commands this WARN_ON triggers
> though... The command name should be printed along with the WARN_ON.

REPLY_RX_PHY_CMD

Stanislaw

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Emmanuel Grumbach]

Emmanuel Grumbach
egrumbach@gmail.com


On Tue, Jul 10, 2012 at 9:02 AM, Stanislaw Gruszka <sgruszka@redhat.com> wrote:
> On Mon, Jul 09, 2012 at 10:10:38PM +0300, Emmanuel Grumbach wrote:
>> I would be glad to know on what commands this WARN_ON triggers
>> though... The command name should be printed along with the WARN_ON.
>
> REPLY_RX_PHY_CMD
>

Ok thanks. This comes from the fw for every HT frame received in 4965.
Do you see that the WARNING triggers for every HT frame Rx?
I don't mind removing that code anyway, I was just asking to
understand what happens :-)

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Stanislaw Gruszka]

On Tue, Jul 10, 2012 at 04:15:09PM +0300, Emmanuel Grumbach wrote:
> Emmanuel Grumbach
> egrumbach@gmail.com
> 
> 
> On Tue, Jul 10, 2012 at 9:02 AM, Stanislaw Gruszka <sgruszka@redhat.com> wrote:
> > On Mon, Jul 09, 2012 at 10:10:38PM +0300, Emmanuel Grumbach wrote:
> >> I would be glad to know on what commands this WARN_ON triggers
> >> though... The command name should be printed along with the WARN_ON.
> >
> > REPLY_RX_PHY_CMD
> >
> 
> Ok thanks. This comes from the fw for every HT frame received in 4965.
> Do you see that the WARNING triggers for every HT frame Rx?
> I don't mind removing that code anyway, I was just asking to
> understand what happens :-)

I did not hit that. It was raported for iwl1000 on RHEL6.3:
https://bugzilla.redhat.com/show_bug.cgi?id=837754

Stanislaw

Re: [RHEL6 PATCH] iwlwifi: remove spurious warning

[Emmanuel Grumbach]

>>
>> On Tue, Jul 10, 2012 at 9:02 AM, Stanislaw Gruszka <sgruszka@redhat.com> wrote:
>> > On Mon, Jul 09, 2012 at 10:10:38PM +0300, Emmanuel Grumbach wrote:
>> >> I would be glad to know on what commands this WARN_ON triggers
>> >> though... The command name should be printed along with the WARN_ON.
>> >
>> > REPLY_RX_PHY_CMD
>> >
>>
>> Ok thanks. This comes from the fw for every HT frame received in 4965.
>> Do you see that the WARNING triggers for every HT frame Rx?
>> I don't mind removing that code anyway, I was just asking to
>> understand what happens :-)
>
> I did not hit that. It was raported for iwl1000 on RHEL6.3:
> https://bugzilla.redhat.com/show_bug.cgi?id=837754
>

Ok, and the bug says - Occasional... Anyway.