authorgravatar for mlugg@mlugg.co.ukMatthew Lugg <mlugg@mlugg.co.uk> 2026-08-13 22:39:32+01:00
committergravatar for mlugg@mlugg.co.ukMatthew Lugg <mlugg@mlugg.co.uk> 2026-08-16 10:37:19+01:00
loga2ddff74f82b518ed7d0b22ff88bf3b1ee97dbb2
tree4b4750135f798ebbb55a8cc3fd8559968d64aa80
parent58c84c59c000c341d7abcf98def0cd13d0e1630e
signaturelock-open Commit is signed but in an unrecognized format.

Elf2: fix incorrect validation in SPARC relocation

It turns out that SPARC has a strange definition of "PC-relative" where the offsets are... unsigned? This results in some annoying restrictions when linking: specifically, the GOT needs to appear after all code segments, because these PC-relative relocations are used to access it. When violating that, we should at least get relocation errors, but because we were doing the wrong validation here, we were just triggering miscompilations instead.

1 files changed, 1 insertions(+), 1 deletions(-)

src/link/Elf2.zig+1-1
......@@ -6665,7 +6665,7 @@ fn addRelocAssumeCapacity(
66656665
66666666 .WDISP30 => try elf.addSymbolRelocAssumeCapacity(node, offset, target, addend, .simple(.rel, .{ .dest = .@"32[29:0]", .cast = .signed, .shift = .@"2_exact" })),
66676667 .WPLT30 => try elf.addSymbolRelocAssumeCapacity(node, offset, target, addend, .simple(.pltrel, .{ .dest = .@"32[29:0]", .cast = .signed, .shift = .@"2_exact" })),
6668 .PC22 => try elf.addSymbolRelocAssumeCapacity(node, offset, target, addend, .simple(.rel, .{ .dest = .@"32[21:0]", .cast = .signed, .shift = .@"10" })),
6668 .PC22 => try elf.addSymbolRelocAssumeCapacity(node, offset, target, addend, .simple(.rel, .{ .dest = .@"32[21:0]", .cast = .unsigned, .shift = .@"10" })),
66696669 .H44 => try elf.addSymbolRelocAssumeCapacity(node, offset, target, addend, .simple(.abs, .{ .dest = .@"32[21:0]", .cast = .unsigned, .shift = .@"22" })),
66706670 .M44 => try elf.addSymbolRelocAssumeCapacity(node, offset, target, addend, .simple(.abs, .{ .dest = .@"32[9:0]", .cast = .trunc, .shift = .@"12" })),
66716671