<html>
<head>
<base href="https://bugs.llvm.org/">
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW - COFF obj2yaml -> yaml2obj round trip does the wrong thing with globals in .bss"
href="https://bugs.llvm.org/show_bug.cgi?id=41836">41836</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>COFF obj2yaml -> yaml2obj round trip does the wrong thing with globals in .bss
</td>
</tr>
<tr>
<th>Product</th>
<td>libraries
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>Windows NT
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>enhancement
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>Object
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>rnk@google.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvm-bugs@lists.llvm.org
</td>
</tr></table>
<p>
<div>
<pre>Seems related to <a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED FIXED - MC-COFF: section.bss always has .size as 0"
href="show_bug.cgi?id=8338">https://bugs.llvm.org/show_bug.cgi?id=8338</a>
Specifically, SizeOfRawData for .bss sections is not preserved, it is zeroed
out. LLD currently contains a lot of yaml input test cases with accidentally
zero-sized globals in .bss because of this.
Consider:
$ cat bss.s
.bss
.global gv_bss
gv_bss:
.long 0
$ llvm-mc bss.s -o bss.obj -filetype=obj
$ obj2yaml bss.obj | yaml2obj -o bss2.obj
$ llvm-readobj -sections bss.obj > bss1.txt
$ llvm-readobj -sections bss2.obj > bss2.txt
$ diff -u bss1.txt bss2.txt
--- bss1.txt 2019-05-10 14:00:19.121866800 -0700
+++ bss2.txt 2019-05-10 14:00:22.056197400 -0700
@@ -1,5 +1,5 @@
-File: bss.obj
+File: bss2.obj
Format: COFF-x86-64
Arch: x86_64
AddressSize: 64bit
@@ -10,7 +10,7 @@
VirtualSize: 0x0
VirtualAddress: 0x0
RawDataSize: 0
- PointerToRawData: 0x8C
+ PointerToRawData: 0x0
PointerToRelocations: 0x0
PointerToLineNumbers: 0x0
RelocationCount: 0
@@ -28,7 +28,7 @@
VirtualSize: 0x0
VirtualAddress: 0x0
RawDataSize: 0
- PointerToRawData: 0x8C
+ PointerToRawData: 0x0
PointerToRelocations: 0x0
PointerToLineNumbers: 0x0
RelocationCount: 0
@@ -45,7 +45,7 @@
Name: .bss (2E 62 73 73 00 00 00 00)
VirtualSize: 0x0
VirtualAddress: 0x0
- RawDataSize: 4
+ RawDataSize: 0
PointerToRawData: 0x0
PointerToRelocations: 0x0
PointerToLineNumbers: 0x0
The .data and .text differences are mostly meaningless, I think. Reducing the
size of the .bss section is an issue because it changes the meaning of the
object pretty significantly: globals in .bss get smaller.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>