std.zig.llvm.bitcode_writer: Fix word byte order on big endian systems.

The bitcode format always uses little endian words. Prior to this commit, a
bitcode file produced on e.g. aarch64_be or s390x would fail to be loaded by
LLVM.
This commit is contained in:
Alex Rønne Petersen 2025-03-04 00:02:09 +01:00 committed by Andrew Kelley
parent 8363b95178
commit 3c924abb69

View file

@ -68,13 +68,13 @@ pub fn BitcodeWriter(comptime types: []const type) type {
in_bits -= n;
if (self.bit_count != 0) return;
try self.buffer.append(self.bit_buffer);
try self.buffer.append(std.mem.nativeToLittle(u32, self.bit_buffer));
self.bit_buffer = 0;
}
// Write 32-bit chunks of input bits
while (in_bits >= 32) {
try self.buffer.append(@truncate(in_buffer));
try self.buffer.append(std.mem.nativeToLittle(u32, @truncate(in_buffer)));
in_buffer >>= 31;
in_buffer >>= 1;
@ -153,7 +153,7 @@ pub fn BitcodeWriter(comptime types: []const type) type {
pub fn alignTo32(self: *BcWriter) Error!void {
if (self.bit_count == 0) return;
try self.buffer.append(self.bit_buffer);
try self.buffer.append(std.mem.nativeToLittle(u32, self.bit_buffer));
self.bit_buffer = 0;
self.bit_count = 0;
}
@ -209,7 +209,7 @@ pub fn BitcodeWriter(comptime types: []const type) type {
try self.bitcode.alignTo32();
// Set the number of words in the block at the start of the block
self.bitcode.buffer.items[self.start] = @truncate(self.bitcode.length() - self.start - 1);
self.bitcode.buffer.items[self.start] = std.mem.nativeToLittle(u32, @truncate(self.bitcode.length() - self.start - 1));
}
pub fn writeUnabbrev(self: *Self, code: u32, values: []const u64) Error!void {