Tools: check for GCC 9, rather than 6 #10618

kjbracey · 2019-05-20T11:03:44Z

Description

Planning to update to supporting GCC 9 as found in GNU Tools for Arm Embedded Processors 9-2019q4-major.

Newer GCC in particular supports ARMv8-M security extensions, and link-time optimisation - two areas of interest.

Should be no code changes required, but need to change the warning in the toolchain script.

Docs update: ARMmbed/mbed-os-5-docs#1073

Pull request type

[ ] Fix
[ ] Refactor
[ ] Target update
[X] Functionality change
[ ] Docs update
[ ] Test update
[ ] Breaking change

Reviewers

@bulislaw

Release Notes

GCC 9 is now the officially-supported compiler version for the GCC_ARM toolchain

ciarmcom · 2019-05-20T13:00:44Z

@kjbracey-arm, thank you for your changes.
@bulislaw @ARMmbed/mbed-os-tools @ARMmbed/mbed-os-maintainers please review.

tools/test/toolchains/api_test.py

tools/toolchains/gcc.py

0xc0170 · 2019-05-20T13:43:12Z

Thanks @mark-edgeworth for quick review!

0xc0170

Initial approval for the update, it has been due.

Docs and CI needs to have this update integrated

cmonr · 2019-05-20T18:38:26Z

@kjbracey-arm Any reason for jumping straight to GCC 8?
The last time I looked at building w/ multiple GCC versions, compile times increased about ~25% w/ GCC8.

(Can dig up older Travis CI builds if needed)

kjbracey · 2019-05-21T09:50:13Z

@kjbracey-arm Any reason for jumping straight to GCC 8?
The last time I looked at building w/ multiple GCC versions, compile times increased about ~25% w/ GCC8.

No specific reason, beyond it being the current version, and not totally new and raw (ARM embedded release was in December, first update release due in June). And a general assumption that newer = better. (I'm optimistic sometimes!)

I don't see any particular merit in choosing an old version to test against, unless we do actually know of problems with GCC 8.

In terms of features, LTO is usable in GCC 7, but I don't know if GCC 7 is sufficient for ARMv8-M CMSE. It probably is, but can anyone confirm?

kjbracey · 2019-05-21T09:51:33Z

Also, I was comparing against IAR and ARM Compiler 6, where both versions we're using are of similar age to the GCC 8 ARM embedded release, so it felt like making it catch up.

0xc0170 · 2019-05-21T12:09:15Z

The last time I looked at building w/ multiple GCC versions, compile times increased about ~25% w/ GCC8.

I've checked Travis, it fetches still 6.x version - it should be updated and we can compare the time to older logs

cmonr · 2019-05-21T15:20:07Z

@0xc0170 @kjbracey-arm
Feel free to review: https://travis-ci.org/cmonr/mbed-os/builds/496577264

Not conclusive by any means, but it was an interesting to observe.

Also, for updating the GCC version, you should only need to change this line: https://github.com/ARMmbed/mbed-os/blob/master/tools/test/travis-ci/functions.sh#L62

jeromecoutant · 2019-06-21T09:31:49Z

Hi

Any update ?

GCC6 cannot compile TFM.
TFM needs GCC7 at least

@Devran01

0xc0170 · 2019-06-24T14:24:59Z

Also, for updating the GCC version, you should only need to change this line: https://github.com/ARMmbed/mbed-os/blob/master/tools/test/travis-ci/functions.sh#L62

@kjbracey-arm Can you update?

What about CI update, tracking issue for this to happen?

kjbracey · 2019-06-24T14:35:24Z

We have a JIRA PREQ: https://jira.arm.com/browse/ISGDEVPREQ-1487 - no tasks created there yet, and not sure what planning status is for the CI guys.

urutva · 2019-07-11T12:47:12Z

@bulislaw @kjbracey-arm @Patater
We have identified a compiler issue with GCC6.3.1 while building secure side. It looks like the compiler is generating a branch instruction before expanding CMSE related instructions which is causing the assembler to fail with "branch out of range". But it happens only with specific pattern of code which involves multiple if statements.
More info in issue #10887.

While analysing the issue I observed that we are using v7-M FPU flags for v8-M as well.
For both GCC and ARMCC, we are setting -mfpu=fpv5-sp-d16 in case of v8-M. I've created a PR #11024. This solves the issue on few targets but not all.

The issue is not seen with GCC7. Can we prioritise this PR and switch to either GCC7 or GCC8?

0xc0170 · 2019-08-14T12:05:42Z

@Devran01 Please update also ticket 1487 referenced above in our internal tracker. I would like to understand what we are doing with this update in the next release or we don't.

urutva · 2019-08-19T11:36:14Z

@0xc0170 I've updated https://jira.arm.com/browse/ISGDEVPREQ-1487 with relavent details.

OPpuolitaival · 2019-08-26T13:08:13Z

GCC version updated in CI. Triggering CI for this.

kjbracey · 2019-10-25T07:08:05Z

I think this can proceed to merge now. CI is on GCC 8, and I'm seeing the warnings in the logs.

0xc0170 · 2019-10-25T11:52:47Z

CI started

mbed-ci · 2019-10-25T15:11:49Z

Test run: SUCCESS

Summary: 11 of 11 test jobs passed
Build number : 4
Build artifacts

0xc0170 · 2019-10-28T08:41:12Z

@OPpuolitaival Just to confirm, this is ready for integration ?

OPpuolitaival · 2019-10-28T09:47:40Z

We can merge this from my point of view

bulislaw · 2019-11-12T11:56:14Z

Does that need to be updated to 9?

0xc0170 · 2019-11-13T17:51:24Z

I would probably wait for 9 to be out at least one more quarter as it was just released. @kjbracey-arm what do you think?

kjbracey · 2019-11-14T10:01:22Z

I have no particularly strong feelings. If specifying 9, would we be open to moving to requiring later update releases based on 9 during LTS? If so, taking the first 9 is probably no biggy.

This should just follow whatever decision is being made - let me know :)

0xc0170 · 2019-11-26T09:19:48Z

@kjbracey-arm This shall be updated to 9 version and CI started to get in 5.15?

kjbracey · 2019-11-26T09:20:24Z

Will do.

Planning to update to supporting GCC 9 as found in GNU Tools for Arm Embedded Processors 9-2019q4-major. Newer GCC in particular supports ARMv8-M security extensions, and link-time optimisation - two areas of interest. Should be no code changes required, but need to change the warning in the toolchain script.

0xc0170 · 2019-11-26T10:05:30Z

CI started

mbed-ci · 2019-11-26T11:32:24Z

Test run: SUCCESS

Summary: 11 of 11 test jobs passed
Build number : 5
Build artifacts

kjbracey force-pushed the gcc8 branch from 508304a to ab7778a Compare May 20, 2019 11:51

kjbracey mentioned this pull request May 20, 2019

GCC version 6 -> 9 ARMmbed/mbed-os-5-docs#1073

Closed

ciarmcom requested review from bulislaw and a team May 20, 2019 13:00

ciarmcom added the needs: review label May 20, 2019

mark-edgeworth suggested changes May 20, 2019

View reviewed changes

tools/test/toolchains/api_test.py Outdated Show resolved Hide resolved

tools/toolchains/gcc.py Outdated Show resolved Hide resolved

mark-edgeworth approved these changes May 20, 2019

View reviewed changes

0xc0170 requested a review from OPpuolitaival May 20, 2019 13:44

0xc0170 approved these changes May 20, 2019

View reviewed changes

0xc0170 added the do not merge label May 23, 2019

bulislaw approved these changes Jun 21, 2019

View reviewed changes

0xc0170 added needs: CI release-version: 5.14.0-rc1 and removed do not merge needs: review labels Aug 26, 2019

0xc0170 added needs: CI and removed needs: work labels Oct 25, 2019

0xc0170 added ready for merge needs: review and removed needs: CI ready for merge labels Oct 28, 2019

0xc0170 added release-version: 5.15.0-rc1 and removed release-version: 6.0.0-alpha-1 First pre-release version of 6.0.0 labels Nov 26, 2019

kjbracey force-pushed the gcc8 branch from ab7778a to 5bb7cc4 Compare November 26, 2019 09:52

kjbracey force-pushed the gcc8 branch from 5bb7cc4 to 5342f0f Compare November 26, 2019 09:52

kjbracey changed the title ~~Tools: check for GCC 8, rather than 6~~ Tools: check for GCC 9, rather than 6 Nov 26, 2019

0xc0170 approved these changes Nov 26, 2019

View reviewed changes

0xc0170 added needs: CI and removed needs: review labels Nov 26, 2019

0xc0170 removed the needs: CI label Nov 26, 2019

0xc0170 merged commit 27012ff into ARMmbed:master Nov 26, 2019

kjbracey mentioned this pull request May 28, 2020

Mbed Cloud Client Example with Mbed OS6 hardfaults #13017

Closed

Tools: check for GCC 9, rather than 6 #10618

Tools: check for GCC 9, rather than 6 #10618

Uh oh!

Conversation

kjbracey commented May 20, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Pull request type

Reviewers

Release Notes

Uh oh!

ciarmcom commented May 20, 2019

Uh oh!

Uh oh!

Uh oh!

0xc0170 commented May 20, 2019

Uh oh!

0xc0170 left a comment

Choose a reason for hiding this comment

Uh oh!

cmonr commented May 20, 2019

Uh oh!

kjbracey commented May 21, 2019

Uh oh!

kjbracey commented May 21, 2019

Uh oh!

0xc0170 commented May 21, 2019

Uh oh!

cmonr commented May 21, 2019

Uh oh!

jeromecoutant commented Jun 21, 2019

Uh oh!

0xc0170 commented Jun 24, 2019

Uh oh!

kjbracey commented Jun 24, 2019

Uh oh!

urutva commented Jul 11, 2019

Uh oh!

0xc0170 commented Aug 14, 2019

Uh oh!

urutva commented Aug 19, 2019

Uh oh!

OPpuolitaival commented Aug 26, 2019

Uh oh!

kjbracey commented Oct 25, 2019

Uh oh!

0xc0170 commented Oct 25, 2019

Uh oh!

mbed-ci commented Oct 25, 2019

Test run: SUCCESS

Uh oh!

0xc0170 commented Oct 28, 2019

Uh oh!

OPpuolitaival commented Oct 28, 2019

Uh oh!

bulislaw commented Nov 12, 2019

Uh oh!

0xc0170 commented Nov 13, 2019

Uh oh!

kjbracey commented Nov 14, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 commented Nov 26, 2019

Uh oh!

kjbracey commented Nov 26, 2019

Uh oh!

0xc0170 commented Nov 26, 2019

Uh oh!

mbed-ci commented Nov 26, 2019

Test run: SUCCESS

Uh oh!

Uh oh!

kjbracey commented May 20, 2019 •

edited

Loading

kjbracey commented Nov 14, 2019 •

edited

Loading