[PATCH -next 0/3] Optimize code for ftrace

[PATCH -next 0/3] Optimize code for ftrace

[Wang Wensheng]

Those are small optimizations discovered by code review.

Wang Wensheng (3):
  ftrace: Fix the possible incorrect kernel message
  ftrace: Optimize the allocation for mcount entries
  ftrace: Delete unused variable ftrace_update_time

 kernel/trace/ftrace.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

-- 
2.17.1

[PATCH -next 1/3] ftrace: Fix the possible incorrect kernel message

[Wang Wensheng]

If the number of mcount entries is an integer multiple of
ENTRIES_PER_PAGE, the page count showing on the console would be wrong.

Signed-off-by: Wang Wensheng <wangwensheng4@huawei.com>
---
 kernel/trace/ftrace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index fbf2543111c0..c571c2813c11 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -7401,7 +7401,7 @@ void __init ftrace_init(void)
 	}
 
 	pr_info("ftrace: allocating %ld entries in %ld pages\n",
-		count, count / ENTRIES_PER_PAGE + 1);
+		count, DIV_ROUND_UP(count, ENTRIES_PER_PAGE));
 
 	ret = ftrace_process_locs(NULL,
 				  __start_mcount_loc,
-- 
2.17.1

[PATCH -next 2/3] ftrace: Optimize the allocation for mcount entries

[Wang Wensheng]

If we can't allocate this size, try something smaller with half of the
size. Its order should be decreased by one instead of divided by two.

Signed-off-by: Wang Wensheng <wangwensheng4@huawei.com>
---
 kernel/trace/ftrace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index c571c2813c11..43a958b28022 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -3200,7 +3200,7 @@ static int ftrace_allocate_records(struct ftrace_page *pg, int count)
 		/* if we can't allocate this size, try something smaller */
 		if (!order)
 			return -ENOMEM;
-		order >>= 1;
+		order--;
 		goto again;
 	}
 
-- 
2.17.1

[PATCH -next 3/3] ftrace: Delete unused variable ftrace_update_time

[Wang Wensheng]

ftrace_update_time is not used anywhere else after initialization. So
delete it together with its initial code.

Signed-off-by: Wang Wensheng <wangwensheng4@huawei.com>
---
 kernel/trace/ftrace.c | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 43a958b28022..3e102c44b117 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -3110,7 +3110,6 @@ int ftrace_shutdown(struct ftrace_ops *ops, int command)
 	return 0;
 }
 
-static u64		ftrace_update_time;
 unsigned long		ftrace_update_tot_cnt;
 unsigned long		ftrace_number_of_pages;
 unsigned long		ftrace_number_of_groups;
@@ -3130,13 +3129,10 @@ static int ftrace_update_code(struct module *mod, struct ftrace_page *new_pgs)
 	bool init_nop = ftrace_need_init_nop();
 	struct ftrace_page *pg;
 	struct dyn_ftrace *p;
-	u64 start, stop;
 	unsigned long update_cnt = 0;
 	unsigned long rec_flags = 0;
 	int i;
 
-	start = ftrace_now(raw_smp_processor_id());
-
 	/*
 	 * When a module is loaded, this function is called to convert
 	 * the calls to mcount in its text to nops, and also to create
@@ -3173,8 +3169,6 @@ static int ftrace_update_code(struct module *mod, struct ftrace_page *new_pgs)
 		}
 	}
 
-	stop = ftrace_now(raw_smp_processor_id());
-	ftrace_update_time = stop - start;
 	ftrace_update_tot_cnt += update_cnt;
 
 	return 0;
-- 
2.17.1

Re: [PATCH -next 3/3] ftrace: Delete unused variable ftrace_update_time

[Steven Rostedt]

On Wed, 9 Nov 2022 09:44:34 +0000
Wang Wensheng <wangwensheng4@huawei.com> wrote:

> ftrace_update_time is not used anywhere else after initialization. So
> delete it together with its initial code.

Thanks for the first two patches. They are definitely fixes and I plan on
testing and adding them to my next pull request.

As for this one, this has been more of a reminder for me to one day (I
know, it's been years!) to export this to see from user space. I sometimes
do it manually with a printk(), but it should probably be always shown.
Perhaps in dyn_ftrace_total_info ?

-- Steve