commit bf285c7e409bddfe7089e570ebf836d0343c8628 (tree)
parent faa1849f81fd1478f2340ab6ccdc8d71d1c5d102
Author: Jakub Konka <kubkon@jakubkonka.com>
Date: Sat, 13 Jan 2024 19:12:55 +0100
test/link/macho: test for correct handling of large __bss sections
Diffstat:
1 file changed, 22 insertions(+), 0 deletions(-)
diff --git a/test/link/macho.zig b/test/link/macho.zig
@@ -10,6 +10,7 @@ pub fn testAll(b: *Build, build_opts: BuildOptions) *Step {
macho_step.dependOn(testDeadStrip(b, .{ .target = default_target }));
macho_step.dependOn(testEntryPointDylib(b, .{ .target = default_target }));
+ macho_step.dependOn(testLargeBss(b, .{ .target = default_target }));
macho_step.dependOn(testMhExecuteHeader(b, .{ .target = default_target }));
macho_step.dependOn(testSectionBoundarySymbols(b, .{ .target = default_target }));
macho_step.dependOn(testSegmentBoundarySymbols(b, .{ .target = default_target }));
@@ -162,6 +163,27 @@ fn testEntryPointDylib(b: *Build, opts: Options) *Step {
return test_step;
}
+fn testLargeBss(b: *Build, opts: Options) *Step {
+ const test_step = addTestStep(b, "macho-large-bss", opts);
+
+ // TODO this test used use a 4GB zerofill section but this actually fails and causes every
+ // linker I tried misbehave in different ways. This only happened on arm64. I thought that
+ // maybe S_GB_ZEROFILL section is an answer to this but it doesn't seem supported by dyld
+ // anymore. When I get some free time I will re-investigate this.
+ const exe = addExecutable(b, opts, .{ .name = "main", .c_source_bytes =
+ \\char arr[0x1000000];
+ \\int main() {
+ \\ return arr[2000];
+ \\}
+ });
+
+ const run = addRunArtifact(exe);
+ run.expectExitCode(0);
+ test_step.dependOn(&run.step);
+
+ return test_step;
+}
+
fn testMhExecuteHeader(b: *Build, opts: Options) *Step {
const test_step = addTestStep(b, "macho-mh-execute-header", opts);