<html>
<head>
<base href="https://llvm.org/bugs/" />
</head>
<body><span class="vcard"><a class="email" href="mailto:ck@remobjects.com" title="Carlo Kok <ck@remobjects.com>"> <span class="fn">Carlo Kok</span></a>
</span> changed
<a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED INVALID - ELF executable generated with lld crashes when reading DYNAMIC link_map on x86-64"
href="https://llvm.org/bugs/show_bug.cgi?id=28040">bug 28040</a>
<br>
<table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>What</th>
<th>Removed</th>
<th>Added</th>
</tr>
<tr>
<td style="text-align:right;">Status</td>
<td>NEW
</td>
<td>RESOLVED
</td>
</tr>
<tr>
<td style="text-align:right;">Resolution</td>
<td>---
</td>
<td>INVALID
</td>
</tr></table>
<p>
<div>
<b><a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED INVALID - ELF executable generated with lld crashes when reading DYNAMIC link_map on x86-64"
href="https://llvm.org/bugs/show_bug.cgi?id=28040#c7">Comment # 7</a>
on <a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED INVALID - ELF executable generated with lld crashes when reading DYNAMIC link_map on x86-64"
href="https://llvm.org/bugs/show_bug.cgi?id=28040">bug 28040</a>
from <span class="vcard"><a class="email" href="mailto:ck@remobjects.com" title="Carlo Kok <ck@remobjects.com>"> <span class="fn">Carlo Kok</span></a>
</span></b>
<pre>So this ended up being a user error. Below code isn't supposed to work at all,
from the glibc mailing list:
<span class="quote">> The l_addr is a *relocation*, or a delta between first PT_LOAD
> .p_vaddr and the actual load address.</span >
<span class="quote">> For non-PIE ET_EXEC, this delta is always 0. For non-prelinked ET_DYN
> (which are normally linked with first PT_LOAD .p_vaddr == 0) it's the
> actual load address (this is why it works most of the time).</span >
<span class="quote">> But for any prelinked ET_DYN binary with first PT_LOAD .p_vaddr != 0,
> this will crash and burn. See also "man prelink".</span >
The reason this hit at all for me is because my compiler didn't emit version
records in the dummy .so files so weak dl_iterate_phdr wouldn't be resolved by
the loader. libGC (boehm) used the code above as a fallback.
closed the issue as invalid.</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>