https://bugs.llvm.org/show_bug.cgi?id=47902

            Bug ID: 47902
           Summary: Assertion failure for thinLTO + --save-temps +
                    non-ascii file name
           Product: lld
           Version: unspecified
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: normal
          Priority: P
         Component: ELF
          Assignee: [email protected]
          Reporter: [email protected]
                CC: [email protected], [email protected]

I was writing a test based on an internal test that made sure the linker could
handle non-ascii filenames when LTO was used. Whilst experimenting with LLD's
--save-temps option, I noticed that the linker will assert when a Thin bitcode
file is passed in, with a non-ascii name (but not when the --save-temps option
is not specified, nor for regular LTO. This is the lit test I was writing:

; REQUIRES: x86

;; Show that both regular LTO and ThinLTO work correctly when an input file
uses
;; a non-ascii filename.

;; Regular LTO.
; RUN: llvm-as %s -o %t£.o
; RUN: ld.lld %t£.o -o %t --save-temps
; RUN: llvm-readelf -s %t | FileCheck %s

;; Thin LTO.
; RUN: opt -module-summary %s -o %t-thin£.o
; RUN: ld.lld %t-thin£.o -o %t-thin --save-temps
; RUN: llvm-readelf -s %t-thin | FileCheck %s

; CHECK: _start

target datalayout =
"e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

@_start = global i32 0

(The llvm-readelf lines probably would want replacing with checks that the
appropriate temporary files have been created). The following assertion is
fired when it gets to the thinLTO link.

Assertion failed: (Val & ~(~0U >> (32-NumBits))) == 0 && "High bits set!", file
C:\llvm\llvm-project\llvm\include\llvm/Bitstream/BitstreamWriter.h, line 204

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs

Reply via email to