CMakeLists.txt: -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR #16536

marc-hb · 2019-06-01T05:27:43Z

In commit 28a5657 we stopped ZEPHYR_BASE from leaking into
__FILE__ and other macros. This works great for apps inside ZEPHYR_BASE
but does nothing for apps outside ZEPHYR_BASE.
-fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR does it.

To avoid collisions and for consistency change:
-fmacro-prefix-map=${ZEPHYR_BASE}=.
to:
-fmacro-prefix-map=${ZEPHYR_BASE}=ZEPHYR_BASE

Quickest test/demo:

Copy samples/hello_word/ to ~/home_world/
Add "%s", __FILE__ to printf in ~/home_world/src/main.c

cmake -B build -S ~/home_world/ -DBOARD=qemu_x86
make -C build run

Before:
~/home_world/src/main.c says Hello World! qemu_x86

After:
CMAKE_SOURCE_DIR/src/main.c says Hello World! qemu_x86

objdump -h $(find build -name *.c.obj) | grep noinit

9 .noinit."ZEPHYR_BASE/kernel/system_work_q.c".0 0000
17 .noinit."ZEPHYR_BASE/kernel/init.c".2 00000100 000
18 .noinit."ZEPHYR_BASE/kernel/init.c".1 00000400 000
19 .noinit."ZEPHYR_BASE/kernel/init.c".3 00000800 000
16 .noinit."ZEPHYR_BASE/kernel/mailbox.c".2 00000348
18 .noinit."ZEPHYR_BASE/kernel/mailbox.c".1 00000028
20 .noinit."ZEPHYR_BASE/kernel/pipes.c".2 00000280 00
22 .noinit."ZEPHYR_BASE/kernel/pipes.c".1 00000028 00

Signed-off-by: Marc Herbert [email protected]

mped-oticon

LGTM

ozhuraki

LGTM

SebastianBoe

Can this create aliasing issues between ZEPHYR_BASE and the app dir?

Could -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=app resolve this?

marc-hb · 2019-06-03T14:11:44Z

Can this create aliasing issues between ZEPHYR_BASE and the app dir?

Can you be more specific?

Could -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=app resolve this?

Why not.

SebastianBoe · 2019-06-03T15:20:25Z

Can this create aliasing issues between ZEPHYR_BASE and the app dir?

Can you be more specific?

For instance, Zephyr has the file zephyr/include/init.h. Imagine that a user has a different file located at app_dir/include/init.h. Then after pre-processing, the path would look like ./include/init.h. Now imagine that, for whatever reason, some tooling needs to be able to deduce whether this is referencing the zephyr or app init.h. It would not be able to, a human would also be confused, but should at least eventually be able to figure it out.

EDIT: If we introduce =app, I suppose we should introduce =zephyr as well to be consistent.

marc-hb · 2019-06-03T20:09:16Z

Thanks @SebastianBoe, makes sense.

I'll just wait a couple days to give people a chance to bikeshed the prefix names app/ and zephyr/.

Shooting first: zephyr_base/?

mbolivar · 2019-06-03T22:52:46Z

Shooting first: zephyr_base/?

ZEPHYR_BASE?

SebastianBoe · 2019-06-04T09:59:33Z

If we go for ZEPHYR_BASE then CMAKE_HOME_DIRECTORY would be the corresponding name for the app.

marc-hb · 2019-06-04T14:10:11Z

I think you meant CMAKE_SOURCE_DIR?

https://cmake.org/cmake/help/latest/variable/CMAKE_HOME_DIRECTORY.html

CMAKE_HOME_DIR should not be used in project code. The variable CMAKE_SOURCE_DIR has the same value and should be preferred.

Not like this would be "using it in project code" for real but it would still be a bad example (I had never heard of CMAKE_HOME_DIR until now)

SebastianBoe · 2019-06-04T14:12:56Z

Right, CMAKE_SOURCE_DIR would be better.

mbolivar · 2019-06-04T19:54:51Z

Random question, but what about zephyr modules? That doesn't seem to be covered in this PR (no problem, incremental progress is good!), but I'm curious what ought to be done while we are thinking about this.

ZEPHYR_MODULE/<mumble-mumble> ?

marc-hb · 2019-06-04T20:30:10Z

Random question, but what about zephyr modules?

Very good and not so random question. I haven't observed any module actually triggering any difference for me yet but I guess a list of -fmacro-prefix-map= options could be dynamically generated from build/zephyr_modules.txt ?

Note this is (at least) the second issue caused by moving modules higher than ZEPHYR_BASE, here's another, cmake one I've been working on: https://cmake.org/pipermail/cmake/2019-June/069547.html

Don't get me wrong: nesting git repositories is horrible so I'm not even thinking about that.

Maybe a new WEST_TOP gradually replacing ZEPHYR_BASE? I think it's been mentioned in unrelated context(s). Just thinking out loud.

In commit 28a5657 we stopped ZEPHYR_BASE from leaking into __FILE__ and other macros. This works great for apps inside ZEPHYR_BASE but does nothing for apps outside ZEPHYR_BASE. -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR does it. To avoid collisions and for consistency change: -fmacro-prefix-map=${ZEPHYR_BASE}=. to: -fmacro-prefix-map=${ZEPHYR_BASE}=ZEPHYR_BASE Quickest test/demo: - Copy samples/hello_word/ to ~/home_world/ - Add "%s", __FILE__ to printf in ~/home_world/src/main.c cmake -B build -S ~/home_world/ -DBOARD=qemu_x86 make -C build run Before: ~/home_world/src/main.c says Hello World! qemu_x86 After: CMAKE_SOURCE_DIR/src/main.c says Hello World! qemu_x86 objdump -h $(find build -name *.c.obj) | grep noinit 9 .noinit."ZEPHYR_BASE/kernel/system_work_q.c".0 0000 17 .noinit."ZEPHYR_BASE/kernel/init.c".2 00000100 000 18 .noinit."ZEPHYR_BASE/kernel/init.c".1 00000400 000 19 .noinit."ZEPHYR_BASE/kernel/init.c".3 00000800 000 16 .noinit."ZEPHYR_BASE/kernel/mailbox.c".2 00000348 18 .noinit."ZEPHYR_BASE/kernel/mailbox.c".1 00000028 20 .noinit."ZEPHYR_BASE/kernel/pipes.c".2 00000280 00 22 .noinit."ZEPHYR_BASE/kernel/pipes.c".1 00000028 00 Signed-off-by: Marc Herbert <[email protected]>

mbolivar · 2019-06-04T22:38:30Z

Maybe a new WEST_TOP gradually replacing ZEPHYR_BASE? I think it's been mentioned in unrelated context(s). Just thinking out loud.

I want to reach for that too, but we need a solution that works without assuming west, for those easterners out there ;).

marc-hb · 2019-06-05T01:03:03Z

https://app.shippable.com/github/zephyrproject-rtos/zephyr/runs/43105/5/console 2h time out, stuck at

JOBS: 12
Selecting default platforms per test case
Building testcase defconfigs...

4 other Shippable shards PASS

Done

marc-hb · 2019-06-11T23:43:48Z

I haven't observed any module actually triggering any difference for me yet

Famous last words.

sanitycheck -T $ZEPHYR_BASE/tests/posix/fs/ -p qemu_x86
readelf --wide --decompress --string-dump=.rodata.disk_status.str1.1 sanity-out/qemu_x86/tests/posix/fs/portability.posix/modules/fatfs/CMakeFiles/..__modules__fs__fatfs.dir/zfs_diskio.c.obj
  [     0]  /home/bob/tmp/repro/westisbest/modules/fs/fatfs/zfs_diskio.c
  [    4c]  pdrv < ((long) (((int) sizeof(char[1 - 2 * !(!__builtin_types_compatible_p(__typeof__(pdrv_str), __typeof__(&(pdrv_str)[0])))]) - 1) + (sizeof(pdrv_str) / sizeof((pdrv_str)[0]))))
  [   100]  ASSERTION FAIL [%s] @ %s:%d^J
  [   11e]  pdrv out-of-range^J^J

Any ASSERT in a module should reproduce

Quick and dirty hack below, moving on for now... bigger fish to fry.

get_filename_component(west_top ${ZEPHYR_BASE} DIRECTORY)
zephyr_cc_option(-fmacro-prefix-map=${west_top}=WEST_TOP)

marc-hb · 2019-06-17T12:57:58Z

Non-urgent ping? 2 lines, 5 approvers :-)

marc-hb · 2019-09-26T19:44:38Z

Maybe a new WEST_TOP gradually replacing ZEPHYR_BASE? I think it's been mentioned in unrelated context(s). Just thinking out loud.

TA-DA! zephyrproject-rtos/west#311

marc-hb · 2019-09-27T21:53:37Z

cmake: -fmacro-prefix-map=${WEST_TOPDIR}=WEST_TOPDIR #19440

marc-hb requested review from nashif and SebastianBoe as code owners June 1, 2019 05:27

marc-hb requested review from andrewboie, galak, mped-oticon and ozhuraki June 1, 2019 18:20

mped-oticon approved these changes Jun 2, 2019

View reviewed changes

nashif approved these changes Jun 2, 2019

View reviewed changes

ozhuraki approved these changes Jun 3, 2019

View reviewed changes

SebastianBoe previously requested changes Jun 3, 2019

View reviewed changes

aescolar added the area: Build System label Jun 3, 2019

marc-hb requested a review from mbolivar June 3, 2019 20:08

marc-hb force-pushed the macro-prefix-source-dir branch from 4371d20 to 135afcc Compare June 4, 2019 22:21

marc-hb changed the title ~~CMakeLists.txt: add -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=.~~ CMakeLists.txt: -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR Jun 4, 2019

marc-hb requested a review from SebastianBoe June 4, 2019 22:41

marc-hb mentioned this pull request Jun 4, 2019

meta issue for reproducible builds (was: tests/determinism.sh) #14593

Closed

marc-hb closed this Jun 5, 2019

marc-hb reopened this Jun 5, 2019

mbolivar approved these changes Jun 11, 2019

View reviewed changes

SebastianBoe approved these changes Jun 12, 2019

View reviewed changes

marc-hb added area: Toolchains Toolchains Enhancement Changes/Updates/Additions to existing features labels Jun 15, 2019

nashif merged commit f67dcdb into zephyrproject-rtos:master Jun 17, 2019

marc-hb deleted the macro-prefix-source-dir branch June 25, 2019 18:53

marc-hb mentioned this pull request Sep 18, 2019

CMake enables -fmacro-prefix-map on GCC 7 #15551

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

CMakeLists.txt: -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR #16536

CMakeLists.txt: -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR #16536

marc-hb commented Jun 1, 2019 •

edited

Loading

mped-oticon left a comment

ozhuraki left a comment

SebastianBoe left a comment

marc-hb commented Jun 3, 2019

SebastianBoe commented Jun 3, 2019 •

edited

Loading

marc-hb commented Jun 3, 2019

mbolivar commented Jun 3, 2019

SebastianBoe commented Jun 4, 2019

marc-hb commented Jun 4, 2019

SebastianBoe commented Jun 4, 2019

mbolivar commented Jun 4, 2019

marc-hb commented Jun 4, 2019

mbolivar commented Jun 4, 2019

marc-hb commented Jun 5, 2019

marc-hb commented Jun 11, 2019 •

edited

Loading

marc-hb commented Jun 17, 2019

marc-hb commented Sep 26, 2019

marc-hb commented Sep 27, 2019

CMakeLists.txt: -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR #16536

CMakeLists.txt: -fmacro-prefix-map=${CMAKE_SOURCE_DIR}=CMAKE_SOURCE_DIR #16536

Conversation

marc-hb commented Jun 1, 2019 • edited Loading

mped-oticon left a comment

Choose a reason for hiding this comment

ozhuraki left a comment

Choose a reason for hiding this comment

SebastianBoe left a comment

Choose a reason for hiding this comment

marc-hb commented Jun 3, 2019

SebastianBoe commented Jun 3, 2019 • edited Loading

marc-hb commented Jun 3, 2019

mbolivar commented Jun 3, 2019

SebastianBoe commented Jun 4, 2019

marc-hb commented Jun 4, 2019

SebastianBoe commented Jun 4, 2019

mbolivar commented Jun 4, 2019

marc-hb commented Jun 4, 2019

mbolivar commented Jun 4, 2019

marc-hb commented Jun 5, 2019

marc-hb commented Jun 11, 2019 • edited Loading

marc-hb commented Jun 17, 2019

marc-hb commented Sep 26, 2019

marc-hb commented Sep 27, 2019

marc-hb commented Jun 1, 2019 •

edited

Loading

SebastianBoe commented Jun 3, 2019 •

edited

Loading

marc-hb commented Jun 11, 2019 •

edited

Loading