[PATCH] tools/nolibc: Fix WIFSIGNALED(0)

[PATCH] tools/nolibc: Fix WIFSIGNALED(0)

[Daniel Palmer]

WIFSIGNALED(0) currently returns true, so a process that exited
normally with status 0 looks like it was killed by a signal.

The macro uses a trick of checking whether status - 1 is below
0xff but is subtly broken because of using signed arithmetic.
For 0 it computes -1 < 0xff which is true.

Do what musl does: mask the status to 16 bits and use unsigned
arithmetic so 0 wraps to 0xffff and the comparison fails.
The mask also prevents bits we don't care about from sneaking in.

Fixes: 8c934d4822c7 ("tools/nolibc: add helpers for wait() signal exits")
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
---

I'm putting off sending anything else (dirfd and some other bits)
until after the merge window has finished.. but this one seems like
it should be fixed sooner.

I found this while debugging an annoying issue in my homebrew nolibc
based shell.

 tools/include/nolibc/types.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/include/nolibc/types.h b/tools/include/nolibc/types.h
index 8f3cb18df7f1..d241629b9e67 100644
--- a/tools/include/nolibc/types.h
+++ b/tools/include/nolibc/types.h
@@ -119,7 +119,7 @@ struct timeval {
 #define WEXITSTATUS(status) (((status) & 0xff00) >> 8)
 #define WIFEXITED(status)   (((status) & 0x7f) == 0)
 #define WTERMSIG(status)    ((status) & 0x7f)
-#define WIFSIGNALED(status) ((status) - 1 < 0xff)
+#define WIFSIGNALED(status) (((status) & 0xffff) - 1U < 0xffU)
 
 /* standard exit() codes */
 #define EXIT_SUCCESS 0
-- 
2.53.0

Re: [PATCH] tools/nolibc: Fix WIFSIGNALED(0)

[Thomas Weißschuh]

Hi Daniel,

On 2026-10-18 19:53:41+0900, Daniel Palmer wrote:
> WIFSIGNALED(0) currently returns true, so a process that exited
> normally with status 0 looks like it was killed by a signal.
> 
> The macro uses a trick of checking whether status - 1 is below
> 0xff but is subtly broken because of using signed arithmetic.
> For 0 it computes -1 < 0xff which is true.
> 
> Do what musl does: mask the status to 16 bits and use unsigned
> arithmetic so 0 wraps to 0xffff and the comparison fails.
> The mask also prevents bits we don't care about from sneaking in.

We already have commit 368fae47c6ca ("tools/nolibc: fix WIFSIGNALED()
for status 0") for this issue in nolibc-next. This implementation is
masking the status with 0xffff. OTOH the musl implementation of
WIFCONTINUED() is also missing the masking, so I am not sure how
necessary it is.

I'll look into it some more.

> Fixes: 8c934d4822c7 ("tools/nolibc: add helpers for wait() signal exits")
> Signed-off-by: Daniel Palmer <daniel@thingy.jp>
> ---
> 
> I'm putting off sending anything else (dirfd and some other bits)
> until after the merge window has finished.. but this one seems like
> it should be fixed sooner.

Feel free to send the patches already. We won't apply them until after
the merge window, but we can discuss them before. There are no other
trees with changes to nolibc, so there won't be any conflicts.

> I found this while debugging an annoying issue in my homebrew nolibc
> based shell.
> 
>  tools/include/nolibc/types.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/tools/include/nolibc/types.h b/tools/include/nolibc/types.h
> index 8f3cb18df7f1..d241629b9e67 100644
> --- a/tools/include/nolibc/types.h
> +++ b/tools/include/nolibc/types.h
> @@ -119,7 +119,7 @@ struct timeval {
>  #define WEXITSTATUS(status) (((status) & 0xff00) >> 8)
>  #define WIFEXITED(status)   (((status) & 0x7f) == 0)
>  #define WTERMSIG(status)    ((status) & 0x7f)
> -#define WIFSIGNALED(status) ((status) - 1 < 0xff)
> +#define WIFSIGNALED(status) (((status) & 0xffff) - 1U < 0xffU)
>  
>  /* standard exit() codes */
>  #define EXIT_SUCCESS 0
> -- 
> 2.53.0
>