| Message ID | 158981614256.106494.12226121528668381542.stgit@bahia.lan (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | scripts/sorttable: Correctly handle mmap() returning MAP_FAILED | expand |
diff --git a/scripts/sorttable.c b/scripts/sorttable.c index ec6b5e81eba1..5ad7a9bbff42 100644 --- a/scripts/sorttable.c +++ b/scripts/sorttable.c @@ -91,6 +91,7 @@ static void *mmap_file(char const *fname, size_t *size) addr = mmap(0, sb.st_size, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0); if (addr == MAP_FAILED) { fprintf(stderr, "Could not mmap file: %s\n", fname); + addr = NULL; goto out; }
The caller of mmap_file() assumes it returns a valid address or NULL on error. If mmap() fails for some reason, MAP_FAILED is returned instead and sorttable crashes later when trying to dereference the pointer: Program received signal SIGSEGV, Segmentation fault. 0x0000000000402b25 in do_file (fname=0x7fffffffe5e2 "vmlinux", addr=0xffffffffffffffff) at scripts/sorttable.c:264 264 switch (ehdr->e_ident[EI_DATA]) { (gdb) p ehdr $1 = (Elf32_Ehdr *) 0xffffffffffffffff mmap() can only return NULL if the user explicitely asks for it with MAP_FIXED, which isn't the case here. So, rather than changing the semantics of mmap_file() and having the caller to cope with an extra sentinel, return NULL when mmap() fails. This bug exists since the addition of the sortextable binary (previous name of sorttable). That code was borrowed from scripts/recordmount.c which had the same issue. It got fixed in a similar manner by commit 3f1df12019f3 ("recordmcount: Rewrite error/success handling"). Cc: stable@vger.kernel.org # v3.5 Fixes: a79f248b9b30 ("scripts: Add sortextable to sort the kernel's exception table.") Signed-off-by: Greg Kurz <groug@kaod.org> --- scripts/sorttable.c | 1 + 1 file changed, 1 insertion(+)