lkml.org 
[lkml]   [2023]   [Jan]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[tip: perf/core] perf/core: Change the layout of perf_sample_data
The following commit has been merged into the perf/core branch of tip:

Commit-ID: 7bdb1767bf011c7f6065ac483ad2f00e434c3979
Gitweb: https://git.kernel.org/tip/7bdb1767bf011c7f6065ac483ad2f00e434c3979
Author: Namhyung Kim <namhyung@kernel.org>
AuthorDate: Thu, 29 Dec 2022 12:40:59 -08:00
Committer: Ingo Molnar <mingo@kernel.org>
CommitterDate: Mon, 09 Jan 2023 12:22:09 +01:00

perf/core: Change the layout of perf_sample_data

The layout of perf_sample_data is designed to minimize cache-line
access. The perf_sample_data_init() used to initialize a couple of
fields unconditionally so they were placed together at the head.

But it's changed now to set the fields according to the actual
sample_type flags. The main user (the perf tools) sets the IP, TID,
TIME, PERIOD always. Also group relevant fields like addr, phys_addr
and data_page_size.

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20221229204101.1099430-1-namhyung@kernel.org
---
include/linux/perf_event.h | 34 +++++++++++++++++++---------------
1 file changed, 19 insertions(+), 15 deletions(-)

diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index ad92ad3..03949d0 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -1098,47 +1098,51 @@ extern u64 perf_event_read_value(struct perf_event *event,

struct perf_sample_data {
/*
- * Fields set by perf_sample_data_init(), group so as to
- * minimize the cachelines touched.
+ * Fields set by perf_sample_data_init() unconditionally,
+ * group so as to minimize the cachelines touched.
*/
u64 sample_flags;
u64 period;

/*
- * The other fields, optionally {set,used} by
- * perf_{prepare,output}_sample().
+ * Fields commonly set by __perf_event_header__init_id(),
+ * group so as to minimize the cachelines touched.
*/
- struct perf_branch_stack *br_stack;
- union perf_sample_weight weight;
- union perf_mem_data_src data_src;
- u64 txn;
- u64 addr;
- struct perf_raw_record *raw;
-
u64 type;
- u64 ip;
struct {
u32 pid;
u32 tid;
} tid_entry;
u64 time;
u64 id;
- u64 stream_id;
struct {
u32 cpu;
u32 reserved;
} cpu_entry;
+
+ /*
+ * The other fields, optionally {set,used} by
+ * perf_{prepare,output}_sample().
+ */
+ u64 ip;
struct perf_callchain_entry *callchain;
- u64 aux_size;
+ struct perf_raw_record *raw;
+ struct perf_branch_stack *br_stack;
+ union perf_sample_weight weight;
+ union perf_mem_data_src data_src;
+ u64 txn;

struct perf_regs regs_user;
struct perf_regs regs_intr;
u64 stack_user_size;

- u64 phys_addr;
+ u64 stream_id;
u64 cgroup;
+ u64 addr;
+ u64 phys_addr;
u64 data_page_size;
u64 code_page_size;
+ u64 aux_size;
} ____cacheline_aligned;

/* default value for data source */
\
 
 \ /
  Last update: 2023-03-26 23:33    [W:0.082 / U:0.228 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site