2018-05-03 21:05:08

by Jerry Hoemann

[permalink] [raw]
Subject: [PATCH 0/1] watchdog/hpwdt: Claim NMIs generated by iLO5

Guenter,

In v4 of my Febuary patch series:

https://lkml.org/lkml/2018/2/25/201

I had a merge error in hpwdt.c that causes hpwdt to not claim
NMIs that it should on iLO5 (i.e. Gen10) systems.

The Feb patches were recently picked up for 4.17-rc2.

Not sure the appropriate action here.

The merge error ocurred on:
https://lkml.org/lkml/2018/2/25/203
And was carried forward through out the remaining patches
in the series.

We could ask for patches to be reverted and reworked. Or,
we can apply a 1-line fix.

If the latter, here is the 1-line fix.

Jerry Hoemann (1):
watchdog: hpwdt: Claim NMIs generated by iLO5

drivers/watchdog/hpwdt.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

--
2.13.6



2018-05-03 21:04:02

by Jerry Hoemann

[permalink] [raw]
Subject: [PATCH 1/1] watchdog: hpwdt: Claim NMIs generated by iLO5

On iLO5 going forward we want to return and not claim the NMI, if
the NMI was NOT gnerated by the iLO as a result of the watchdog
timing out or an explicit generate NMI.

The sense of the test in is inverted and prevents hpwdt_pretimeout
from claiming NMIs when it should.

Signed-off-by: Jerry Hoemann <[email protected]>
---
drivers/watchdog/hpwdt.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/watchdog/hpwdt.c b/drivers/watchdog/hpwdt.c
index a43ab2cecca2..9dc62a461451 100644
--- a/drivers/watchdog/hpwdt.c
+++ b/drivers/watchdog/hpwdt.c
@@ -159,7 +159,7 @@ static int hpwdt_pretimeout(unsigned int ulReason, struct pt_regs *regs)
"3. OA Forward Progress Log\n"
"4. iLO Event Log";

- if (ilo5 && ulReason == NMI_UNKNOWN && mynmi)
+ if (ilo5 && ulReason == NMI_UNKNOWN && !mynmi)
return NMI_DONE;

if (ilo5 && !pretimeout)
--
2.13.6


2018-05-03 21:21:10

by Guenter Roeck

[permalink] [raw]
Subject: Re: [PATCH 1/1] watchdog: hpwdt: Claim NMIs generated by iLO5

On Thu, May 03, 2018 at 03:00:55PM -0600, Jerry Hoemann wrote:
> On iLO5 going forward we want to return and not claim the NMI, if
> the NMI was NOT gnerated by the iLO as a result of the watchdog
> timing out or an explicit generate NMI.
>
> The sense of the test in is inverted and prevents hpwdt_pretimeout
> from claiming NMIs when it should.
>
> Signed-off-by: Jerry Hoemann <[email protected]>

Fixes: a042229a18ac ("watchdog: hpwdt: Update nmi_panic message.")
Reviewed-by: Guenter Roeck <[email protected]>

> ---
> drivers/watchdog/hpwdt.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/watchdog/hpwdt.c b/drivers/watchdog/hpwdt.c
> index a43ab2cecca2..9dc62a461451 100644
> --- a/drivers/watchdog/hpwdt.c
> +++ b/drivers/watchdog/hpwdt.c
> @@ -159,7 +159,7 @@ static int hpwdt_pretimeout(unsigned int ulReason, struct pt_regs *regs)
> "3. OA Forward Progress Log\n"
> "4. iLO Event Log";
>
> - if (ilo5 && ulReason == NMI_UNKNOWN && mynmi)
> + if (ilo5 && ulReason == NMI_UNKNOWN && !mynmi)
> return NMI_DONE;
>
> if (ilo5 && !pretimeout)
> --
> 2.13.6
>