linux-kernel - Re: [PATCH v8 7/9] perf/amd/iommu: Check return value when set and get counter value

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sun, 22 Jan 2017 20:53:45 +0100
From:   Borislav Petkov <bp@...en8.de>
To:     Suravee Suthikulpanit <Suravee.Suthikulpanit@....com>
Cc:     linux-kernel@...r.kernel.org, iommu@...ts.linux-foundation.org,
        joro@...tes.org, peterz@...radead.org, mingo@...hat.com
Subject: Re: [PATCH v8 7/9] perf/amd/iommu: Check return value when set and
 get counter value

On Mon, Jan 16, 2017 at 01:23:34AM -0600, Suravee Suthikulpanit wrote:
> From: Suravee Suthikulpanit <suravee.suthikulpanit@....com>
> 
> In, perf_iommu_start(), we need to check the return value from
> amd_iommu_set_reg(). In case of failure, we should not enable the PMU.
> 
> Also, in perf_iommu_read(), we need to check the return value from
> amd_iommu_get_reg() before using the value.

So basically you wanna say:

"Add amd_iommu_{g,s}et_reg() error handling."

> 
> Cc: Peter Zijlstra <peterz@...radead.org>
> Cc: Borislav Petkov <bp@...en8.de>
> Cc: Joerg Roedel <joro@...tes.org>
> Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@....com>
> ---
>  arch/x86/events/amd/iommu.c | 19 +++++++++++--------
>  1 file changed, 11 insertions(+), 8 deletions(-)
> 
> diff --git a/arch/x86/events/amd/iommu.c b/arch/x86/events/amd/iommu.c
> index 200d2e8..cc7bea4 100644
> --- a/arch/x86/events/amd/iommu.c
> +++ b/arch/x86/events/amd/iommu.c
> @@ -284,6 +284,7 @@ static void perf_iommu_disable_event(struct perf_event *event)
>  
>  static void perf_iommu_start(struct perf_event *event, int flags)
>  {
> +	u64 val;
>  	struct hw_perf_event *hwc = &event->hw;
>  
>  	if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
> @@ -292,15 +293,16 @@ static void perf_iommu_start(struct perf_event *event, int flags)
>  	WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
>  	hwc->state = 0;
>  
> -	if (flags & PERF_EF_RELOAD) {
> -		u64 prev_raw_count =  local64_read(&hwc->prev_count);
> -		amd_iommu_pc_set_reg(0, _GET_BANK(event), _GET_CNTR(event),
> -				     IOMMU_PC_COUNTER_REG, &prev_raw_count);
> -	}
> +	if (!(flags & PERF_EF_RELOAD))
> +		return;

This looks wrong - you still need to do

	perf_iommu_enable_event(event);
	perf_event_update_userpage(event);

even if !PERF_EF_RELOAD.

> +	val = local64_read(&hwc->prev_count) & GENMASK_ULL(48, 0);
> +	if (amd_iommu_pc_set_reg(0, _GET_BANK(event), _GET_CNTR(event),
> +				   IOMMU_PC_COUNTER_REG, &val))
> +		return;
>  
>  	perf_iommu_enable_event(event);
>  	perf_event_update_userpage(event);
> -
>  }

-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.