Adding case insensitive 'error' detection

[bridadan]

Description

GCC Assembler errors were being missed because it prints 'error' with a captial 'E'. This change allows the 'e' to be either lower case or upper case.

Status

READY

Migrations

If this PR changes any APIs or behaviors, give a short description of what API users should do when this PR is merged.

NO

Related PRs

This may conflict with #3653

Todos

• Tests

Notes

Before, the tools would miss errors like the following when it was printed from the assembler:

/tmp/ccXLitIR.s: Assembler messages:
/tmp/ccXLitIR.s:677: Error: cannot honor width suffix -- `mov r1,#0'
/tmp/ccXLitIR.s:679: Error: cannot honor width suffix -- `mov r0,#0'
/tmp/ccXLitIR.s:727: Error: cannot honor width suffix -- `mov r1,#4'
/tmp/ccXLitIR.s:729: Error: cannot honor width suffix -- `neg r1,r1'
/tmp/ccXLitIR.s:731: Error: cannot honor width suffix -- `mov r5,#4'
/tmp/ccXLitIR.s:737: Error: cannot honor width suffix -- `mov r1,#3'
/tmp/ccXLitIR.s:739: Error: cannot honor width suffix -- `neg r1,r1'
/tmp/ccXLitIR.s:741: Error: cannot honor width suffix -- `mov r5,#3'
/tmp/ccXLitIR.s:743: Error: cannot honor width suffix -- `neg r5,r5'
/tmp/ccXLitIR.s:754: Error: cannot honor width suffix -- `orr r5,r2'
/tmp/ccXLitIR.s:904: Error: cannot honor width suffix -- `mov r1,#4'
/tmp/ccXLitIR.s:906: Error: cannot honor width suffix -- `neg r1,r1'
/tmp/ccXLitIR.s:908: Error: cannot honor width suffix -- `mov r0,#4'
/tmp/ccXLitIR.s:915: Error: cannot honor width suffix -- `mov r1,#3'
/tmp/ccXLitIR.s:916: Error: cannot honor width suffix -- `neg r1,r1'
/tmp/ccXLitIR.s:919: Error: cannot honor width suffix -- `mov r0,#3'
/tmp/ccXLitIR.s:921: Error: cannot honor width suffix -- `neg r0,r0'
/tmp/ccXLitIR.s:926: Error: lo register required -- `add r0,r0,#12'
/tmp/ccXLitIR.s:982: Error: cannot honor width suffix -- `mov r1,#6'
/tmp/ccXLitIR.s:984: Error: cannot honor width suffix -- `neg r1,r1'
/tmp/ccXLitIR.s:986: Error: cannot honor width suffix -- `mov r0,#6'
/tmp/ccXLitIR.s:987: Error: cannot honor width suffix -- `neg r0,r0'

Now it should list every one of the errors.

[theotherjimmy]

lol XD

[bulislaw]

How about making it all case insensitive?

[bridadan]

@bulislaw I thought about it, but I wasn't sure what impact that'd have. I'd rather not make too many changes to this sensitive line at the moment.

[bulislaw]

maybe at least [Ww] for warnings :)

[bridadan]

I thought about that too :) But I don't know if the assembler even outputs warnings that way, have you seen that?

[bulislaw]

Ok, lets take a step back :) I don't really mind about all of that, but it's shame to add 1-2 days delay while chasing down issues like that, ultimately my goal is to be able to understand what's going on just by looking at CI logs. I don't understand why we just can't keep full logs, how much storage would it take? And i don't mean -v but just full stdout/stderr output from the compiler.

But yes I agree the simplest step we can do is to merge this change, which solves my problem for now.

[0xc0170]

/morph test

[mbed-bot]

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1487

All builds and test passed!

[bridadan]

add 1-2 days delay while chasing down issues like that, ultimately my goal is to be able to understand what's going on just by looking at CI logs

I agree, I've added verbose logs to the CI 'Build' step now, so hopefully this won't block others in the future.

But also if you notice locally that an error is missing without the -v, please don't forget to raise an issue here!

[bulislaw]

Thanks!

[sg-]

@bridadan needs an update!

[bridadan]

/morph test

[bridadan]

retest uvisor

[mbed-bot]

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1497

All builds and test passed!

[adbridge]

@mazimkhan

[mazimkhan]

retest uvisor