linux-kernel - Re: [PATCH V6 06/13] perf: Add Nehalem and Sandy Bridge uncore support

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]
Message-ID: <CABPqkBQYRs9F2U3LVWcxip3-2ijxAE8+szj2TY8MKwO2w7se1A@mail.gmail.com>
Date:	Mon, 18 Jun 2012 11:23:29 +0200
From:	Stephane Eranian <eranian@...gle.com>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
Cc:	"Yan, Zheng" <zheng.z.yan@...el.com>, mingo@...e.hu,
	jolsa@...hat.com, andi@...stfloor.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH V6 06/13] perf: Add Nehalem and Sandy Bridge uncore support

On Sat, Jun 16, 2012 at 2:46 PM, Peter Zijlstra <a.p.zijlstra@...llo.nl> wrote:
>
> On Fri, 2012-06-15 at 20:46 +0200, Stephane Eranian wrote:
> > Yes, and let the parser be case insensitive.
>
> Right, I did the below on top..
>
> ---
> --- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
> @@ -179,22 +179,17 @@ static struct attribute *snbep_uncore_pc
>  };
>
>  static struct uncore_event_desc snbep_uncore_imc_events[] = {
> -       INTEL_UNCORE_EVENT_DESC(CLOCKTICKS, "config=0xffff"),
> -       /* read */
> -       INTEL_UNCORE_EVENT_DESC(CAS_COUNT_RD, "event=0x4,umask=0x3"),
> -       /* write */
> -       INTEL_UNCORE_EVENT_DESC(CAS_COUNT_WR, "event=0x4,umask=0xc"),
> +       INTEL_UNCORE_EVENT_DESC(clockticks,      "event=0xff,umask=0xff");

Just like for core, I would pick 0xff as the pseudo event code for
fixed uncore events
and use the umask value to determine which of the fixed events we want
to map to. There
is only one fixed counter so far. But more could be added later on, so
why not start with
umask=0x00 or umask=0x1 instead of umask=0xff?

> +       INTEL_UNCORE_EVENT_DESC(cas_count_read,  "event=0x04,umask=0x03"),
> +       INTEL_UNCORE_EVENT_DESC(cas_count_write, "event=0x04,umask=0x0c"),
>        { /* end: all zeroes */ },
>  };
>
>  static struct uncore_event_desc snbep_uncore_qpi_events[] = {
> -       INTEL_UNCORE_EVENT_DESC(CLOCKTICKS, "event=0x14"),
> -       /* outgoing data+nondata flits */
> -       INTEL_UNCORE_EVENT_DESC(TxL_FLITS_ACTIVE, "event=0x0,umask=0x6"),
> -       /* DRS data received */
> -       INTEL_UNCORE_EVENT_DESC(DRS_DATA, "event=0x2,umask=0x8"),
> -       /* NCB data received */
> -       INTEL_UNCORE_EVENT_DESC(NCB_DATA, "event=0x3,umask=0x4"),
> +       INTEL_UNCORE_EVENT_DESC(clockticks,       "event=0x14"),
> +       INTEL_UNCORE_EVENT_DESC(txl_flits_active, "event=0x0,umask=0x6"),
> +       INTEL_UNCORE_EVENT_DESC(drs_data,         "event=0x2,umask=0x8"),
> +       INTEL_UNCORE_EVENT_DESC(ncb_data,         "event=0x3,umask=0x4"),
>        { /* end: all zeroes */ },
>  };
>
> @@ -621,29 +616,15 @@ static struct attribute_group nhm_uncore
>  };
>
>  static struct uncore_event_desc nhm_uncore_events[] = {
> -       INTEL_UNCORE_EVENT_DESC(CLOCKTICKS, "config=0xffff"),
> -       /* full cache line writes to DRAM */
> -       INTEL_UNCORE_EVENT_DESC(QMC_WRITES_FULL_ANY,
> "event=0x2f,umask=0xf"),
> -       /* Quickpath Memory Controller normal priority read requests */
> -       INTEL_UNCORE_EVENT_DESC(QMC_NORMAL_READS_ANY,
> "event=0x2c,umask=0xf"),
> -       /* Quickpath Home Logic read requests from the IOH */
> -       INTEL_UNCORE_EVENT_DESC(QHL_REQUEST_IOH_READS,
> -                               "event=0x20,umask=0x1"),
> -       /* Quickpath Home Logic write requests from the IOH */
> -       INTEL_UNCORE_EVENT_DESC(QHL_REQUEST_IOH_WRITES,
> -                               "event=0x20,umask=0x2"),
> -       /* Quickpath Home Logic read requests from a remote socket */
> -       INTEL_UNCORE_EVENT_DESC(QHL_REQUEST_REMOTE_READS,
> -                               "event=0x20,umask=0x4"),
> -       /* Quickpath Home Logic write requests from a remote socket */
> -       INTEL_UNCORE_EVENT_DESC(QHL_REQUEST_REMOTE_WRITES,
> -                               "event=0x20,umask=0x8"),
> -       /* Quickpath Home Logic read requests from the local socket */
> -       INTEL_UNCORE_EVENT_DESC(QHL_REQUEST_LOCAL_READS,
> -                               "event=0x20,umask=0x10"),
> -       /* Quickpath Home Logic write requests from the local socket */
> -       INTEL_UNCORE_EVENT_DESC(QHL_REQUEST_LOCAL_WRITES,
> -                               "event=0x20,umask=0x20"),
> +       INTEL_UNCORE_EVENT_DESC(clockticks,
>  "event=0xff,umask=0xff"),
> +       INTEL_UNCORE_EVENT_DESC(qmc_writes_full_any,
> "event=0x2f,umask=0x0f"),
> +       INTEL_UNCORE_EVENT_DESC(qmc_normal_reads_any,
>  "event=0x2c,umask=0x0f"),
> +       INTEL_UNCORE_EVENT_DESC(qhl_request_ioh_reads,
> "event=0x20,umask=0x01"),
> +       INTEL_UNCORE_EVENT_DESC(qhl_request_ioh_writes,
>  "event=0x20,umask=0x02"),
> +       INTEL_UNCORE_EVENT_DESC(qhl_request_remote_reads,
>  "event=0x20,umask=0x04"),
> +       INTEL_UNCORE_EVENT_DESC(qhl_request_remote_writes,
> "event=0x20,umask=0x08"),
> +       INTEL_UNCORE_EVENT_DESC(qhl_request_local_reads,
> "event=0x20,umask=0x10"),
> +       INTEL_UNCORE_EVENT_DESC(qhl_request_local_writes,
>  "event=0x20,umask=0x20"),
>        { /* end: all zeroes */ },
>  };
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/