<table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Issue</th>
<td>
<a href=https://github.com/llvm/llvm-project/issues/54682>54682</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>
Misoptimization (possibly in MemCpyOpt) results in no output
</td>
</tr>
<tr>
<th>Labels</th>
<td>
llvm:optimizations
</td>
</tr>
<tr>
<th>Assignees</th>
<td>
</td>
</tr>
<tr>
<th>Reporter</th>
<td>
ormris
</td>
</tr>
</table>
<pre>
The following test case produces no output at O2 when "B" is expected. We've bisected this issue to commit f5446b769a7929806f72256fccd4826d66502e59:
```
commit f5446b769a7929806f72256fccd4826d66502e59
Author: Olle Fredriksson [fredriksson.olle@gmail.com](mailto:fredriksson.olle@gmail.com)
Date: Wed Apr 21 22:48:28 2021 +0200
[MemCpyOpt] Allow variable lengths in memcpy optimizer
This makes the memcpy-memcpy and memcpy-memset optimizations work for
variable sizes as long as they are equal, relaxing the old restriction
that they are constant integers. If they're not equal, the old
requirement that they are constant integers with certain size
restrictions is used.
The implementation works by pushing the length tests further down in the
code, which reveals some places where it's enough that the lengths are
equal (but not necessarily constant).
Differential Revision: https://reviews.llvm.org/D100870
```
Steps to reproduce:
```
$ cat test.cpp
#include <string.h>
#include <stdio.h>
void testcase(const char *path, char *name)
{
if (name)
{
*name = 0;
}
char tmp[260];
int pos = (int)strlen(path) - 1;
int pos2 = sizeof(tmp) - 1;
bool loop = true;
while (pos >= 0 && pos2 >= 0 && loop)
{
switch (path[pos])
{
case '\\':
case '/':
memcpy(name, tmp + pos2 + 1, sizeof(tmp) - pos2 - 1);
name[sizeof(tmp) - pos2 - 1] = 0;
loop = false;
break;
default:
tmp[pos2--] = path[pos--];
break;
}
}
}
int main(void)
{
char name[260];
testcase("A/B", name);
printf("%s\n", name);
return 1;
}
$ clang -O2 test.cpp
$ ./a.out # "B" is the expected output
$ clang -O1 test.cpp
$ ./a.out
B
$
```
A couple of things I noticed:
1. Moving tmp to global scope seems to workaround this issue.
2. A zero length memcpy is possible, if the input begins with a "/".
Thanks in advance for any help on this.
</pre>
<img width="1px" height="1px" alt="" src="http://email.email.llvm.org/o/eJyVVktv4zYQ_jXyZRBDpvw8-GDHG2APiwXaBXqmpJHFRhJVkrI3_fX9SFuyHGfT1lAUkUPO45tnqvO37Y-SqdBVpc-qOZJj6yiTlqk1Ou8yttRo0p1rO0fS0XdB55IbioTY44-UJf7ZcuY4n9IfHInViSlVNuyQK0FX1nZMTlOm61o5Khbz-TJdLTdytRGbdbwsVkIslkWW5fO1WObL5SIWvNhEyS6KD1Hcv5fx9QlL-t_srtd2nSu1AXP6XlVML4Zzo16t1TBqsS9uyylA4WgeH2upqimkRYtDJNZ-5TTuf3pUDPIO0rGXRoAnp11rSMxICGzN13iJNYkYO5HYxyKOxxZT-EGpb1w_t2_fWwcNaOddRSdplEyhf8XN0ZVAuaGa66x9I906Vau_2Tzy-uH9UctXeNXB7ZcLT9d7sslHO5Zdz0k6pRtLZ21eESlmzHBQw0KgJWmp0ogiGfiDpWHivzpZReKZDFfyZ4gxiNZVjg3rjMo89zFPVyLOhusZRDvZOBjo-MjGTulrEciINdAb7W4irpzH3AyIynDNjfs3znRWrqSMjZNA01t0z2jQ1sc0dRYh_xHETKpuqyAxIBeAs5S-UdvZsgfg4reQb5aKzmDPUK7PjfckFmOemc7Zm3cuVVZCkRPLypLVNbK0kj5HkZOwSDlggoRsdHcsB2uHEIHRY64BNcTdOkVqexgbBisLj1ZvAzgI5A-MPKiigMDGKXD4jU_Keh8iyEvnWusTV7zggaaKz3ZaVad6qs0RW4dZHK9X8YcpfXn_7ri1vlwYvpagzwtBJOYoWC4gOc3att9NVJNVXc4UJc_ec81xWkbJl4_JudJjaniftMoDV18OAVOAhLJSGoC2a6UrvU_6dSNrHrI-Wu17uFThIb6jIqdvdOovQ5EDxVGyvx06DKXOC3F1i1ogYDfq0O0YopdabcN1SFLBZzAYXsfyouaGnmj2eEeESz7QdYGzXsDD0VTrCkmt23DWmY5HRAQkct-LCQp8CSZgvcTTC7jf9Jx-BYRF_iG-e60Xe3AINXczAmt8gS6dCkEfLZ79g49k9yHdx-N7Il3L3eCfZ4-xr8RX3fEx87uPCAW6Bwq6Jft7poHXYv_JJVTx986-_AacCyQ4P5BTw_L13W7Ohewq92DaJVq8yKenXuAN1rD3H_iPYnD4HD58FKHh-TDzufJR9IfAvQLyELmj3MIo4WuGHyk84H263M62SGBXXE5GYoG4eG5-cbi_Yth1phkF880CXzIqiUr8hHnmfeGY0xSqyCmGHoRAMh51fDXtx53rUHRXmUZ8Z5_xvWztB8on5XCHUtyhnZD2XQ9VzNJXX61Vxvm7wkg0m9I3fQotBoGMGnqsdIoSbTPdokcz2rrf9Q1JGt014wntXZkXU9oRhgjdd6rrlIDjiB-r0PQ9-Cr0YhQUPx-mfFTNtY1KCu5B2ok7zj9K2byGcUXmJ9lkfvg0mD3eqOSqJd0ElaaTfJvkm2QjJ065irfflB1PI9ea49V487xuExIyDZ0aGRFkDLPrpDPV9r47HaFml16mtRffoq7_ntB2_oSLsQzAWHws5su1mJTb1WaTy7nMNrxerdYJ52uR56nIC06S2XJdTCqZcmW3CHcYHpgmu7sxyqOyOEzUFlOfiHFLoCPONlMwXq3EfJYuYky0SYJpksM02TfPidkG7dLuaEGsFAaHG1ECi2PDHCSDvwxT7lab2ig7CXZsgxH_AIbdgm4">