mem: fix UB in readInt/writeInt and delete variants

[jacobly0]

In Zig, loads and stores affect @sizeOf(T) bytes of memory. The old implementations of these functions were pointer casting a buffer of @divExact(@bitSizeOf(T), 8) bytes which is not guaranteed to be large enough to perform a load or store of T directly. This only ever worked because llvm has weird store size rules, and in fact, after this change, some non-llvm backends mysteriously started passing these tests.

By making use of inline, the complicated nest of variant functions could all be reduced to four functions without compromising functionality. After removing some incorrect uses of two of these, there were no references to them remaining, so I deleted them too.

---

Migration guide

These apply to both the functions in std.mem and the methods on std.io.GenericReader, std.io.AnyReader, and std.io.Writer.

• readInt(...) → readInt(...)
• readIntLittle(...) → readInt(..., .little)
• readIntBig(...) → std.mem.readInt(..., .big)
• readIntNative(...) → std.mem.readInt(..., @import("builtin").cpu.arch.endian()) or just @as(T, @bitCast(buffer.*)) for the std.mem version
• readIntForeign(...) → @byteSwap(std.mem.readInt(..., @import("builtin").cpu.arch.endian())) or just @byteSwap(@as(T, @bitCast(buffer.*))) for the std.mem version
• writeInt(...) → std.mem.writeInt(...)
• writeIntLittle(...) → std.mem.writeInt(..., .little)
• writeIntBig(...) → std.mem.writeInt(..., .big)
• writeIntNative(...) → std.mem.writeInt(..., @import("builtin").cpu.arch.endian()) or just buffer.* = @bitCast(value) for the std.mem version
• writeIntForeign(...) → std.mem.writeInt(..., @byteSwap(value), @import("builtin").cpu.arch.endian()) or just buffer.* = @bitCast(@byteSwap(value)) for the std.mem version

---

@Luukdegram unfortunately, this new pattern regressed the self-hosted wasm backend, so just letting you know that there are three new .stage2_wasm checks in lib/test_runner.zig, test/behavior/bugs/1851.zig, and test/behavior/struct.zig.

[andrewrk]

These warnings are new when bootstrapping:

[10/13] Building C object CMakeFiles/zig2.dir/zig2.c.o
/home/andy/Downloads/zig/build-release/zig2.c: In function ‘value_Value_readFromMemory__22826’:
/home/andy/Downloads/zig/build-release/zig2.c:1503723: warning: ‘memcpy’ reading 16 bytes from a region of size 10 [-Wstringop-overread]
1503723 |      memcpy(&t47, &t46, sizeof(zig_u128));
        | 
/home/andy/Downloads/zig/build-release/zig2.c:1503582: note: source object ‘t46’ of size 10
1503582 |  uint8_t t46[10];
        | 
/home/andy/Downloads/zig/build-release/zig2.c: In function ‘mem_readPackedIntLittle__anon_233795__233795’:
/home/andy/Downloads/zig/build-release/zig2.c:2732095: warning: ‘memcpy’ reading 16 bytes from a region of size 10 [-Wstringop-overread]
2732095 |  memcpy(&t11, &t10, sizeof(zig_u128));
        | 
/home/andy/Downloads/zig/build-release/zig2.c:2732072: note: source object ‘t10’ of size 10
2732072 |  uint8_t t10[10];
        | 
/home/andy/Downloads/zig/build-release/zig2.c: In function ‘mem_readPackedIntBig__anon_233796__233796’:
/home/andy/Downloads/zig/build-release/zig2.c:2732169: warning: ‘memcpy’ reading 16 bytes from a region of size 10 [-Wstringop-overread]
2732169 |  memcpy(&t13, &t12, sizeof(zig_u128));
        | 
/home/andy/Downloads/zig/build-release/zig2.c:2732131: note: source object ‘t12’ of size 10
2732131 |  uint8_t t12[10];
        | 
[11/13] Building stage3

[squeek502]

Changes to resinator look correct.

[andrewrk]

D:\a\zig\zig\lib\zig.h(2183): warning C4244: 'function': conversion from 'uint16_t' to 'uint8_t', possible loss of data
...
D:\a\zig\zig\lib\zig.h(3366): warning C4391: 'zig_f80 __fmax(zig_f80,zig_f80,zig_f80)': incorrect return type for intrinsic function, expected 'double'
D:\a\zig\zig\lib\zig.h(3366): warning C4392: 'zig_f80 __fmax(zig_f80,zig_f80,zig_f80)': incorrect number of arguments for intrinsic function, expected '2' arguments
test-x86_64-windows-msvc.c(53486): warning C4133: 'function': incompatible types - from 'A__9111 *' to 'A__9107 *'
test-x86_64-windows-msvc.c(53488): warning C4133: 'function': incompatible types - from 'A__9112 *' to 'A__9107 *'
test-x86_64-windows-msvc.c(65402): warning C4047: '=': 'void (__cdecl *)(void)' differs in levels of indirection from 'uint8_t (*)[4]'
test-x86_64-windows-msvc.c(123940): error C2167: '__fmax': too many actual parameters for intrinsic function
test-x86_64-windows-msvc.c(124357): error C2167: '__fmax': too many actual parameters for intrinsic function
test-x86_64-windows-msvc.c(263186): warning C4090: '=': different 'const' qualifiers
D:\a\zig\zig\lib\zig.h(2183): warning C4244: 'function': conversion from 'uint16_t' to 'uint8_t', possible loss of data
...
D:\a\zig\zig\lib\zig.h(3366): warning C4391: 'zig_f80 __fmax(zig_f80,zig_f80,zig_f80)': incorrect return type for intrinsic function, expected 'double'
D:\a\zig\zig\lib\zig.h(3366): warning C4392: 'zig_f80 __fmax(zig_f80,zig_f80,zig_f80)': incorrect number of arguments for intrinsic function, expected '2' arguments
compiler_rt-x86_64-windows-msvc.c(490): warning C4392: 'zig_f80 __fmax(zig_f80,zig_f80,zig_f80)': incorrect number of arguments for intrinsic function, expected '2' arguments
compiler_rt-x86_64-windows-msvc.c(19292): warning C4392: 'zig_f80 __fmax(const zig_f80,const zig_f80,const zig_f80)': incorrect number of arguments for intrinsic function, expected '2' arguments
compiler_rt-x86_64-windows-msvc.c(19292): error C2169: '__fmax': intrinsic function, cannot be defined

huh. I don't understand how that passed the CI before, or why it's passing on x86_64-windows-debug.

[jacobly0]

These warnings are new when bootstrapping:

Have a fix, but waiting for a CI run to complete.

huh. I don't understand how that passed the CI before, or why it's passing on x86_64-windows-debug.

It's obvious what happened (clash between f-max and fma-x), but like you said, who knows why it happened on this run in particular. I can only assume either a bad roll gave us a different image (different compiler version?) somehow, or an msvc bugs means the error is not consistent.

[jacobly0]

Well, not related to the changes in this PR at least: https://github.com/ziglang/zig/actions/runs/6710563026/job/18236031654