Fix std.c.EXC.MASK struct to match definition in mach/exception_types.h by danielchasehooper · Pull Request #21964 · ziglang/zig

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

danielchasehooper · 2024-11-11T15:45:43Z

This change fixes two issues:

The Zig definition of std.c.EXC.MASK had the mask flags starting at bit 0, but the header it's copying has this:

#defineEXC_BAD_ACCESS 1
...
/* * Masks for exception definitions, above * bit zero is unused, therefore 1 word = 31 exception types */#defineEXC_MASK_BAD_ACCESS (1 << EXC_BAD_ACCESS)

the structure wasn't padded to 32 bits

Obviously the mask isn't used anywhere is zig, because it would've caused a packed structure padding compile error. On the chance that it is used in the future, either by the compiler, or as an api provided by std, I figured I'd save future poor souls from mysterious bugs caused by the mask flags bits being off by one bit.

alexrp

Just some naming nitpicks; for unused/padding fields in packed structs that mirror C enums, we prefer to use the bit position as the name.

Otherwise LGTM.

Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

Turns out this was already fixed in ziglang#21964. I have no idea why GitHub showed an incorrect diff in ziglang#21273, or how applying the diff to master was even possible, but here we are.

fix mask struct to match definition in mach/exception_types.h
da64eba

alexrp requested changes Nov 12, 2024
View reviewed changes

Comment threadlib/std/c/darwin.zig Outdated
Comment threadlib/std/c/darwin.zig Outdated

Apply suggestions from code review
c2d81ac
Co-authored-by: Alex Rønne Petersen <alex@alexrp.com>

alexrp approved these changes Nov 12, 2024
View reviewed changes

alexrp enabled auto-merge (squash) November 12, 2024 21:31

alexrp merged commit 9996f8b into ziglang:masterNov 13, 2024

danielchasehooper deleted the mach_exc_mask branch November 13, 2024 14:24

PauloCampana mentioned this pull request Nov 17, 2024
Compiler errors in std pt. 2 #21094
Open

alexrp added a commit that referenced this pull request Jan 27, 2025
Revertd714cf3(#21273)
6196452
Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.

alexrp mentioned this pull request Jan 27, 2025
Revert d714cf33405486d9aa707e9aee8f103a011d06e9 (#21273) #22623
Merged

andrewrk pushed a commit that referenced this pull request Jan 27, 2025
Revertd714cf3(#21273)
5647dd8
Turns out this was already fixed in #21964. I have no idea why GitHub showed an incorrect diff in #21273, or how applying the diff to master was even possible, but here we are.