public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Jaswinder Singh Rajput <jaswinder@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>, Thomas Gleixner <tglx@linutronix.de>,
	x86 maintainers <x86@kernel.org>,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: [git-pull -tip] x86:perf_counter cleanup
Date: Sun, 08 Mar 2009 11:32:39 +0100	[thread overview]
Message-ID: <1236508359.22914.3643.camel@twins> (raw)
In-Reply-To: <1236507232.28127.2.camel@localhost.localdomain>

On Sun, 2009-03-08 at 15:43 +0530, Jaswinder Singh Rajput wrote:
>  
>  static struct pmc_x86_ops *pmc_amd_init(void)
>  {
> -       u64 old;
> -       int bits;
> -
>         nr_counters_generic = 4;
>         nr_counters_fixed = 0;
>         counter_value_mask = 0x0000FFFFFFFFFFFFULL;



I would rather we try something like the below:

diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c
index f585371..ccb4f79 100644
--- a/arch/x86/kernel/cpu/perf_counter.c
+++ b/arch/x86/kernel/cpu/perf_counter.c
@@ -957,12 +957,39 @@ static struct pmc_x86_ops *pmc_amd_init(void)
 	u64 old;
 	int bits;
 
+	pr_info("AMD Performance Monitoring support detected.\n");
+
 	nr_counters_generic = 4;
 	nr_counters_fixed = 0;
-	counter_value_mask = 0x0000FFFFFFFFFFFFULL;
-	counter_value_bits = 48;
+	counter_value_mask = ~0ULL;
 
-	pr_info("AMD Performance Monitoring support detected.\n");
+	rdmsrl(MSR_K7_PERFCTR0, old);
+	wrmsrl(MSR_K7_PERFCTR0, counter_value_mask);
+
+	/*
+	 * Ensure the counter_value_mask write, is completed before
+	 * we try to read it back.
+	 */ 
+	mb();
+
+	/* read the truncated mask */
+	rdmsrl(MSR_K7_PERFCTR0, counter_value_mask);
+	wrmsrl(MSR_K7_PERFCTR0, old);
+
+	bits = 32 + fls(counter_value_mask >> 32);
+	if (bits == 32)
+		bits = fls((u32)counter_value_mask);
+	counter_value_bits = bits;
+
+	if (counter_value_bits != 48 ||
+	    counter_value_mask != ((1ULL << counter_value_bits) - 1)) {
+		printk(KERN_ERR "Bad cpu! speculated through full memory barrier!\n");
+		printk(KERN_ERR " ... old value: %llx\n", old);
+		printk(KERN_ERR " ... new value: %llx\n", counter_value_mask);
+
+		counter_value_bits = 48;
+		counter_value_mask = 0x0000FFFFFFFFFFFFULL;
+	}
 
 	return &pmc_amd_ops;
 }


  parent reply	other threads:[~2009-03-08 10:33 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-03-08 10:13 [git-pull -tip] x86:perf_counter cleanup Jaswinder Singh Rajput
2009-03-08 10:20 ` Peter Zijlstra
2009-03-08 10:39   ` Jaswinder Singh Rajput
2009-03-08 10:41     ` Ingo Molnar
2009-03-08 11:39       ` Jaswinder Singh Rajput
2009-03-08 10:46     ` Peter Zijlstra
2009-03-08 10:32 ` Peter Zijlstra [this message]
2009-03-08 10:34 ` Peter Zijlstra
2009-03-08 15:52   ` [tip:perfcounters/core] x86: perf_counter cleanup Peter Zijlstra

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1236508359.22914.3643.camel@twins \
    --to=peterz@infradead.org \
    --cc=jaswinder@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox