Attempt to fix build #9087 on macOs intel. #9170
Conversation
Disable long 64 bits support and disable zend_string_equal_val assembly implementation then as instructions are not longer understood.
|
FYI: macOS 10 CI should work again in ~2 hours: https://github.blog/changelog/2022-07-20-github-actions-the-macos-10-15-actions-runner-image-is-being-deprecated-and-will-be-removed-by-8-30-22/ |
| @@ -23,7 +23,7 @@ | |||
| #include <stdint.h> | |||
|
|
|||
| /* This is the heart of the whole int64 enablement in zval. */ | |||
| #if defined(__x86_64__) || defined(__LP64__) || defined(_LP64) || defined(_WIN64) | |||
| #if !defined(__APPLE__) && (defined(__x86_64__) || defined(__LP64__) || defined(_LP64) || defined(_WIN64)) | |||
There was a problem hiding this comment.
Ugh, it would be pretty sad, if we couldn't have 64bit integers on macOS.
There was a problem hiding this comment.
Yes I know, not too "proud" about it :-) ... but either we "ignore" the failed test if we re sure it works well in fact in real hardware, mac Intel is a dying trend but will take a certain time to actually die.
There was a problem hiding this comment.
I'd assume the vast majority of people use a 64-bit build of PHP and might thus implicitly be depending on 64-bit integers. @devnexen Just disabling zend_string_equal_val for macOS doesn't do it?
There was a problem hiding this comment.
oh for sure it builds if I do what you say but ... the main "issue" still remain then
There was a problem hiding this comment.
Hmm, what is the diff? In #9087 it is:
sscanf 64-bit signed int '9223372036854775807' (2^63)-1 = 9223372036854775807
sscanf 64-bit unsign int '18446744073709551615' (2^64)-1 = 18446744073709551615
printf 64-bit signed int '9223372036854775807' (2^63)-1 = 9223372036854775807
007+ printf 64-bit signed int '18446744073709551615' (2^64)-1 = 9223372036854775808
007- printf 64-bit signed int '18446744073709551615' (2^64)-1 = 0
DoneThat doesn't look like a general 64bit integer issue; and frankly, that part of the test looks fishy anyway.
There was a problem hiding this comment.
anyhow
but either we "ignore" the failed test
@zeriyoshi said the test pass w/o issue in his intel mac.
There was a problem hiding this comment.
Do we support ARM for mac? Does it work without Rosetta?
It's not listed in https://wiki.php.net/platforms at least.
There was a problem hiding this comment.
The point is that I would like to understand why this test fails on some platforms, and also, why it passes on (most) others. It doesn't appear to be related to the size of integers, since 18446744073709551615 is supposed to be parsed as float anyway.
There was a problem hiding this comment.
Do we support ARM for mac? Does it work without Rosetta? It's not listed in https://wiki.php.net/platforms at least.
It does pass w/o issue on pure arm mode then my fix is not appropriate indeed.
Disable long 64 bits support and disable zend_string_equal_val assembly
implementation then as instructions are not longer understood.