2017-06-23 21:52:19

by Sudip Mukherjee

[permalink] [raw]
Subject: [PATCH] MIPS: perf: fix build failure

The build of gpr_defconfig is failing with the error:
arch/mips/kernel/perf_event_mipsxx.c:
In function 'mipsxx_pmu_map_raw_event':
arch/mips/kernel/perf_event_mipsxx.c:1614:2:
error: duplicate case value

Patch - f7a31b5e7874 ("MIPS: perf: Remove incorrect odd/even counter
handling for I6400") removed the previous case and added it as a
separate case. But patch dd71e57bacb5 ("MIPS: perf: add I6500 handling")
added it back to its old location and thus giving us two duplicate case.

Fixes: dd71e57bacb5 ("MIPS: perf: add I6500 handling")
Signed-off-by: Sudip Mukherjee <[email protected]>
---

The build log is available at:
https://travis-ci.org/sudipm-mukherjee/parport/jobs/246092909

arch/mips/kernel/perf_event_mipsxx.c | 1 -
1 file changed, 1 deletion(-)

diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
index 4b93cc0..733b612 100644
--- a/arch/mips/kernel/perf_event_mipsxx.c
+++ b/arch/mips/kernel/perf_event_mipsxx.c
@@ -1597,7 +1597,6 @@ static const struct mips_perf_event *mipsxx_pmu_map_raw_event(u64 config)
break;
case CPU_P5600:
case CPU_P6600:
- case CPU_I6400:
case CPU_I6500:
/* 8-bit event numbers */
raw_id = config & 0x1ff;
--
1.9.1


2017-06-26 06:44:57

by Marcin Nowakowski

[permalink] [raw]
Subject: Re: [PATCH] MIPS: perf: fix build failure

Hi Sudip,

This patch fixes the build error, but leaves the I6500 handling incorrect.
I had explained to Ralf how the build should be fixed a while ago so
hopefully he will fix it up in his -next branch (dd71e57bacb5 should
have been applied on top of f7a31b5e7874, but in Ralf's tree
f7a31b5e7874 is applied on v4.12-rc4 while dd71e57bacb5 is applied on
v4.12-rc2).
You can see what those commits should look like by examining patchwork
submissions (links available in commit descriptions).

Marcin

On 23.06.2017 23:52, Sudip Mukherjee wrote:
> The build of gpr_defconfig is failing with the error:
> arch/mips/kernel/perf_event_mipsxx.c:
> In function 'mipsxx_pmu_map_raw_event':
> arch/mips/kernel/perf_event_mipsxx.c:1614:2:
> error: duplicate case value
>
> Patch - f7a31b5e7874 ("MIPS: perf: Remove incorrect odd/even counter
> handling for I6400") removed the previous case and added it as a
> separate case. But patch dd71e57bacb5 ("MIPS: perf: add I6500 handling")
> added it back to its old location and thus giving us two duplicate case.
>
> Fixes: dd71e57bacb5 ("MIPS: perf: add I6500 handling")
> Signed-off-by: Sudip Mukherjee <[email protected]>
> ---
>
> The build log is available at:
> https://travis-ci.org/sudipm-mukherjee/parport/jobs/246092909
>
> arch/mips/kernel/perf_event_mipsxx.c | 1 -
> 1 file changed, 1 deletion(-)
>
> diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
> index 4b93cc0..733b612 100644
> --- a/arch/mips/kernel/perf_event_mipsxx.c
> +++ b/arch/mips/kernel/perf_event_mipsxx.c
> @@ -1597,7 +1597,6 @@ static const struct mips_perf_event *mipsxx_pmu_map_raw_event(u64 config)
> break;
> case CPU_P5600:
> case CPU_P6600:
> - case CPU_I6400:
> case CPU_I6500:
> /* 8-bit event numbers */
> raw_id = config & 0x1ff;
>

2017-06-26 21:32:34

by Sudip Mukherjee

[permalink] [raw]
Subject: Re: [PATCH] MIPS: perf: fix build failure

On Mon, Jun 26, 2017 at 08:44:48AM +0200, Marcin Nowakowski wrote:
> Hi Sudip,
>
> This patch fixes the build error, but leaves the I6500 handling incorrect.

I am failing to understand why I6500 handling is incorrect. I am seeing
that after applying my patch the I6500 handling is same as introduced
by dd71e57bacb5.

> I had explained to Ralf how the build should be fixed a while ago so
> hopefully he will fix it up in his -next branch (dd71e57bacb5 should
> have been applied on top of f7a31b5e7874, but in Ralf's tree
> f7a31b5e7874 is applied on v4.12-rc4 while dd71e57bacb5 is applied
> on v4.12-rc2).

Ohh. ok. that explains. But then it should not have applied cleanly.

--
Regards
Sudip