Re: [PATCH V2 1/4] perf: Add hardware breakpoint address mask

[Jacob Shin <jacob.shin@amd.com>]

On Tue, Apr 23, 2013 at 03:18:44PM +0200, Oleg Nesterov wrote:
> On 04/23, Jacob Shin wrote:
> >
> > +__weak int arch_validate_hwbkpt_addr_mask(struct perf_event *bp)
> > +{
> > +	return bp->attr.bp_addr_mask == 0;
> > +}
> > +
> >  static int validate_hw_breakpoint(struct perf_event *bp)
> >  {
> >  	int ret;
> > @@ -393,6 +398,10 @@ static int validate_hw_breakpoint(struct perf_event *bp)
> >  	if (ret)
> >  		return ret;
> >  
> > +	ret = arch_validate_hwbkpt_addr_mask(bp);
> > +	if (ret)
> > +		return ret;
> 
> Well, this looks obviously wrong?
> 
> arch_validate_hwbkpt_addr_mask() fails if bp_addr_mask == 0? and returns
> "1" as the error code.
> 
> Either it should returns something like "bp_addr_mask ? -ENOTSUPP : 0"
> or the caller should do "if (!validate_hw_breakpoint()) return -ERR".

You are absolutely right. I have mixed up "does this arch support address
masks?" vs "is this a valid address mask?" So sorry about that.

Here is the corrected patch:

>From f2054e8af979f024aa2da93f80646db18492479e Mon Sep 17 00:00:00 2001
From: Jacob Shin <jacob.shin@amd.com>
Date: Mon, 22 Apr 2013 17:02:37 -0500
Subject: [PATCH 1/4] perf: Add hardware breakpoint address mask

Some architectures (for us, AMD Family 16h) allow for "don't care" bit
mask to further qualify a hardware breakpoint address, in order to
trap on range of addresses. Update perf uapi to add bp_addr_mask field.

Signed-off-by: Jacob Shin <jacob.shin@amd.com>
---
 include/uapi/linux/perf_event.h |    5 ++++-
 kernel/events/hw_breakpoint.c   |    9 +++++++++
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
index fb104e5..e22e1d1 100644
--- a/include/uapi/linux/perf_event.h
+++ b/include/uapi/linux/perf_event.h
@@ -286,7 +286,10 @@ struct perf_event_attr {
 		__u64		config1; /* extension of config */
 	};
 	union {
-		__u64		bp_len;
+		struct {
+			__u32		bp_len;
+			__u32		bp_addr_mask;
+		};
 		__u64		config2; /* extension of config1 */
 	};
 	__u64	branch_sample_type; /* enum perf_branch_sample_type */
diff --git a/kernel/events/hw_breakpoint.c b/kernel/events/hw_breakpoint.c
index a64f8ae..dde8273 100644
--- a/kernel/events/hw_breakpoint.c
+++ b/kernel/events/hw_breakpoint.c
@@ -385,6 +385,11 @@ int dbg_release_bp_slot(struct perf_event *bp)
 	return 0;
 }
 
+__weak int arch_validate_hwbkpt_addr_mask(struct perf_event *bp)
+{
+	return bp->attr.bp_addr_mask ? -EOPNOTSUPP : 0;
+}
+
 static int validate_hw_breakpoint(struct perf_event *bp)
 {
 	int ret;
@@ -393,6 +398,10 @@ static int validate_hw_breakpoint(struct perf_event *bp)
 	if (ret)
 		return ret;
 
+	ret = arch_validate_hwbkpt_addr_mask(bp);
+	if (ret)
+		return ret;
+
 	if (arch_check_bp_in_kernelspace(bp)) {
 		if (bp->attr.exclude_kernel)
 			return -EINVAL;
-- 
1.7.9.5