thakis added a comment.
This doesn't seem to be working very well:
thakis@thakis:~/src/llvm-project$ cat test.cc
#include <inttypes.h>
#include <stdio.h>
#include <stdint.h>
int main() {
uint16_t hextets[8];
int chars_scanned;
char buf[] = "1234:5678:9abc:def0:1234:5678:9abc:def0";
sscanf(buf,
"%4" SCNx16 ":%4" SCNx16 ":%4" SCNx16 ":%4" SCNx16 ":%4" SCNx16
":%4" SCNx16 ":%4" SCNx16 ":%4" SCNx16 "%n",
&hextets[0], &hextets[1], &hextets[2], &hextets[3], &hextets[4],
&hextets[5], &hextets[6], &hextets[7], &chars_scanned);
for (int i = 0; i < 8; ++i)
printf("%x ", hextets[i]);
printf("%d\n", chars_scanned);
}
thakis@thakis:~/src/llvm-project$ out/gn/bin/clang test.cc -Wall
test.cc:9:3: warning: 'sscanf' may overflow; destination buffer in argument 9
has size 4, but the corresponding field width plus NUL byte is 5
[-Wfortify-source]
sscanf(buf,
^
test.cc:9:3: warning: 'sscanf' may overflow; destination buffer in argument
10 has size 2, but the corresponding field width plus NUL byte is 5
[-Wfortify-source]
2 warnings generated.
thakis@thakis:~/src/llvm-project$ ./a.out
1234 5678 9abc def0 1234 5678 9abc def0 39
1. The warning is emitted twice, but doesn't point at code the 2nd time round
2. That code looks correct to me (ie there shouldn't be any warnings), maybe
`%n` isn't handled correctly?
3. The diag points at the start of the scanf instead of at the faulty arg.
Especially 2 is breaking builds, so I'll revert this for now. Looks like a cool
warning though, looking forward to the relanding :)
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D111833/new/
https://reviews.llvm.org/D111833
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
