Conversation
Mips has a different way of handling NaN. This makes sure the tests pass on MIPS as well.
| 0x7F, 0xF8, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, | ||
| 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xF8, 0x7F | ||
| // Mips processors use a different NaN. | ||
| ])) || buffer.equals(new Uint8Array([ |
There was a problem hiding this comment.
Can we detect running on mips and only assert the alternative value if so?
The performance jit logger is not implemented on mips.
|
Comment addressd. I also added another commit to fix the flag issue. |
| 0x7F, 0xF7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, | ||
| 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xF7, 0x7F | ||
| ]))); | ||
| if (process.arch === 'mips') { |
There was a problem hiding this comment.
According to the docs for process.arch :
The current possible values are:
'arm','arm64','ia32','mips','mipsel','ppc','ppc64','s390','s390x','x32', and'x64'.
so we probably need to be checking for mipsel as well. Maybe mips64el too. Perhaps @kapouer or anyone else with access to mips can check?
There was a problem hiding this comment.
Hm, actually, it seems like we either have to use the first approach or we can just check one byte.
https://sourceware.org/binutils/docs/as/MIPS-NaN-Encodings.html
There was a problem hiding this comment.
If we go back to the first approach, maybe add a comment Allow both quiet and signalling NaNs?
There was a problem hiding this comment.
I have not built nodejs 10 on mipsel yet, but nodejs 8.11 gives the same result as mips on mipsel.
There was a problem hiding this comment.
I check one byte now and the check will then verify if the whole entry is correct and added a comment.
There was a problem hiding this comment.
So now there is no need to be specific about any architecture anymore.
| assert.ok( | ||
| buffer.equals(new Uint8Array( | ||
| [ 0x7F, 0xC0, 0x00, 0x00, 0x00, 0x00, 0xC0, 0x7F ]))); | ||
| } |
There was a problem hiding this comment.
buffer[1] === 0xF7 isn't true for either assert?
BridgeAR
added
the
author ready
Mips has a different way of handling NaN. This makes sure the tests pass on MIPS as well. PR-URL: nodejs#20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The performance jit logger is not implemented on mips. PR-URL: nodejs#20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Mips has a different way of handling NaN. This makes sure the tests pass on MIPS as well. PR-URL: #20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The performance jit logger is not implemented on mips. PR-URL: #20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Mips has a different way of handling NaN. This makes sure the tests pass on MIPS as well. PR-URL: #20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The performance jit logger is not implemented on mips. PR-URL: #20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Mips has a different way of handling NaN. This makes sure the tests pass on MIPS as well. PR-URL: #20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The performance jit logger is not implemented on mips. PR-URL: #20377 Reviewed-By: Richard Lau <riclau@uk.ibm.com>
4 participants
Mips has a different way of handling NaN. This makes sure the tests
pass on MIPS as well.
This partially fix-es #20336 (the - is there to prevent github to close the issue).
Checklist
make -j4 test(UNIX), orvcbuild test(Windows) passes