Replace runtime strip_path function with compiler intrinsic equivalents #6377

ghost · 2018-03-15T15:52:58Z

Description

Replace runtime strip_path function with compiler intrinsic equivalents

Sleep manager tracing strips the path from filenames and uses the result as an identifier to track drivers that unlock/lock sleep tracing. Replace the function that strips the path from the string, replace this function with a new macro, __FILENAME__ which performs the same action in a compiler specific manner.

GCC_ARM, use __builtin_strrchr which is optimized out at compile time.
ARM, use __MODULE__ which returns the filename without a path.
IAR, specify the --no_path_in_file_macros compiler flag.

Change the driver identifier in the sleep tracing manager to a pointer of the stored file path.

Currently, the identifier is 15-byte character array that is copied from the __FILE__ of the driver being profiled. Given that the content of __FILE__ is stored inside the binary for the duration of the application it would be more efficient to store a pointer to this character string. This saves space in the statistics struct and saves time during lookup and insertion.

Pull request type

@deepikabhavnani @SenRamakri @bulislaw

Sleep manager tracing strips the path from filenames and uses the result as an identifier to track drivers that unlock/lock sleep tracing. Replace the function that strips the path from the string, replace this function with a new macro, __FILENAME__ which performs the same action in a compiler specific manner. - GCC_ARM, use __builtin_strrchr which is optimized out at compile time. - ARM, use __MODULE__ which returns the filename without path. - IAR, specifiy the --no_path_in_file_macros compiler flag.

geky · 2018-03-15T16:47:30Z

platform/mbed_power_mgmt.h

+#if defined(__CC_ARM) || (defined(__ARMCC_VERSION) && (__ARMCC_VERSION >= 6010050))
+#define __FILENAME__ __MODULE__
+#elif defined(__GNUC__)
+#define __FILENAME__ (__builtin_strrchr(__FILE__, '/') ? __builtin_strrchr(__FILE__, '/') + 1 : __builtin_strrchr(__FILE__, '\\') ? __builtin_strrchr(__FILE__, '\\') + 1 : __FILE__)


Doesn't GCC map strchr to __builtin_strrchr automatically?

Apparently yes, I just hadn't turned on optimizations when checking, it doesn't for -O0.

You can use __BASE_FILE__ for GNU. https://gcc.gnu.org/onlinedocs/cpp/Common-Predefined-Macros.html
__BASE_FILE__
This macro expands to the name of the main input file, in the form of a C string constant. This is the source file that was specified on the command line of the preprocessor or C compiler.

I tried it, but it doesn't work in this situation, it expands to the full path of the module that is passed into make/compiler so the output is the same as __FILE__ in this scenario, and something completely different in others.

…filepath. The use of __FILE__ macro to get a usable identifier from the driver path causes the path of the file to be stored in the .text region of the binary. Given that this remains for the entire duration of the program, storing a pointer to this string as an identifier is more efficient than copying the contents of the string during lookup/insertion.

deepikabhavnani · 2018-03-15T18:16:13Z

platform/mbed_power_mgmt.h

-        sleep_manager_lock_deep_sleep_internal(); \
-        sleep_tracker_lock(__FILE__, __LINE__);   \
-    } while (0);
+#if defined(__CC_ARM) || (defined(__ARMCC_VERSION) && (__ARMCC_VERSION >= 6010050))


I guess we don't have to check version of ARM compiler here. __CC_ARM should be sufficient

deepikabhavnani

Looks good to me

0xc0170

Looks fine, but the change --no_path_in_file_macros in the compiler flags for IAR - is this required for this patch to land?

Changing the compiler flags sets the minor version label for this

ghost · 2018-03-19T12:11:17Z

I wouldn't say it's necessary, but removing it does change the behaviour between GCC/ARM and IAR.

Sleep locks held:
[id: Timer.cpp, count: 1]
[id: Ticker.h, count: 1]

// IAR
Sleep locks held:
[id: .\mbed-os\drivers\Timer.cpp, count: 1]
[id: D:\work\mbed-os-example-profiling\mbed-os\drivers/Ticker.h, count: 1]

0xc0170 · 2018-03-19T12:47:17Z

Good snippet to see how it changes.

If compiler flag change can be send separately, this bugfix can go into patch release. Otherwise this will be labeled for the next minor release due to changing compiler flags.
@scartmell-arm Please advice

ghost · 2018-03-19T13:23:58Z

That's fine, I've removed it so it can be added elsewhere.

0xc0170 · 2018-03-19T13:31:33Z

platform/mbed_power_mgmt.h

-        sleep_tracker_lock(__FILE__, __LINE__);   \
-    } while (0);
+#if defined(__CC_ARM)
+#define __FILENAME__ __MODULE__


Why is this starting with __ ? This resembles C lib macro but it is not ? As this is in platform header, I expect this goes to the user space (not internal use only).

If it is our own macro, should not use reserved characters (prefix ___). Similar to what we define in mbed toolchain (shouldn't this be there rather?)

0xc0170 · 2018-03-20T14:25:44Z

platform/mbed_toolchain.h

+// string literal.
+#ifndef FILENAME
+#if defined(__CC_ARM)
+#define FILENAME __MODULE__


should it follow the same prefix MBED_FILENAME?

- Move macro definition to mbed_toolchain.h - Remove double underscores from macro which are reserved. - Fix macro for IAR until compiler flags to disable path are added again.

bulislaw · 2018-03-20T15:22:57Z

hal/mbed_sleep_manager.c

    }

    core_util_atomic_incr_u8(&stat->count, 1);

-    debug("LOCK: %s, ln: %i, lock count: %u\r\n", stripped_path, line, deep_sleep_lock);
+    debug("LOCK: %s, ln: %i, lock count: %u\r\n", filename, line, deep_sleep_lock);


Do we really need the lock/unlock calls? I would say they just repeat the same info we get out of the who's holding sleep lock prints.

Those changes are in a different PR: #6349

cmonr · 2018-03-21T23:19:34Z

/morph build

mbed-ci · 2018-03-22T00:00:07Z

Build : SUCCESS

Build number : 1523
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/6377/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build
/morph mbed2-build

mbed-ci · 2018-03-22T06:08:44Z

Exporter Build : FAILURE

Build number : 1163
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/6377/

0xc0170 · 2018-03-22T08:23:12Z

/morph export-build

mbed-ci · 2018-03-22T10:26:25Z

Exporter Build : SUCCESS

Build number : 1166
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/6377/

mbed-ci · 2018-03-22T15:52:32Z

Test : FAILURE

Build number : 1301
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/6377/1301

cmonr · 2018-03-22T16:40:02Z

Hmm, looks like a UBLOX_EVK_ODIN_W2 timeouts.

/morph test

mbed-ci · 2018-03-22T22:22:20Z

Test : SUCCESS

Build number : 1306
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/6377/1306

geky reviewed Mar 15, 2018

View reviewed changes

deepikabhavnani reviewed Mar 15, 2018

View reviewed changes

Remove superfluous compiler check in macro

1e3050a

deepikabhavnani previously approved these changes Mar 16, 2018

View reviewed changes

0xc0170 added the needs: review label Mar 19, 2018

0xc0170 reviewed Mar 19, 2018

View reviewed changes

Remove IAR compiler flag change from PR

6d7576a

ghost dismissed deepikabhavnani’s stale review via 6d7576a March 19, 2018 13:23

0xc0170 reviewed Mar 19, 2018

View reviewed changes

0xc0170 reviewed Mar 20, 2018

View reviewed changes

Fix issues with __FILENAME__ macro

ab2abcb

- Move macro definition to mbed_toolchain.h - Remove double underscores from macro which are reserved. - Fix macro for IAR until compiler flags to disable path are added again.

0xc0170 requested a review from bulislaw March 20, 2018 15:18

bulislaw reviewed Mar 20, 2018

View reviewed changes

bulislaw approved these changes Mar 20, 2018

View reviewed changes

cmonr added needs: CI and removed needs: review labels Mar 21, 2018

cmonr added the ready for merge label Mar 22, 2018

cmonr added release-version: 5.8.1 and removed needs: CI labels Mar 22, 2018

cmonr merged commit ddf70f1 into ARMmbed:master Mar 23, 2018

cmonr removed the ready for merge label Mar 23, 2018

ghost deleted the feature-deep-sleep-tracing-filename-fix branch December 5, 2018 16:48

Replace runtime strip_path function with compiler intrinsic equivalents #6377

Replace runtime strip_path function with compiler intrinsic equivalents #6377

Uh oh!

Conversation

ghost commented Mar 15, 2018 • edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Replace runtime strip_path function with compiler intrinsic equivalents

Change the driver identifier in the sleep tracing manager to a pointer of the stored file path.

Pull request type

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghost Mar 15, 2018 • edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghost Mar 15, 2018 • edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

deepikabhavnani Mar 15, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

deepikabhavnani left a comment

Choose a reason for hiding this comment

Uh oh!

0xc0170 left a comment

Choose a reason for hiding this comment

Uh oh!

ghost commented Mar 19, 2018 • edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 commented Mar 19, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ghost commented Mar 19, 2018

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghost Mar 20, 2018 • edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cmonr commented Mar 21, 2018

Uh oh!

mbed-ci commented Mar 22, 2018

Triggering tests

Uh oh!

mbed-ci commented Mar 22, 2018

Uh oh!

0xc0170 commented Mar 22, 2018

Uh oh!

mbed-ci commented Mar 22, 2018

Uh oh!

mbed-ci commented Mar 22, 2018

Uh oh!

cmonr commented Mar 22, 2018

Uh oh!

mbed-ci commented Mar 22, 2018

Uh oh!

Uh oh!

ghost commented Mar 15, 2018 •

edited by ghost

Loading

ghost Mar 15, 2018 •

edited by ghost

Loading

ghost Mar 15, 2018 •

edited by ghost

Loading

deepikabhavnani Mar 15, 2018 •

edited

Loading

ghost commented Mar 19, 2018 •

edited by ghost

Loading

0xc0170 commented Mar 19, 2018 •

edited

Loading

ghost Mar 20, 2018 •

edited by ghost

Loading