On Tue, Aug 20, 2024 at 01:15:05AM +0200, Stefano Brivio wrote:
> musl, as of 1.2.5, and glibc < 2.34 don't ship a (trivial)
> close_range() implementation. This will probably be added to musl
> soon, by the way:
>   https://www.openwall.com/lists/musl/2024/08/01/9
> 
> Add a weakly-aliased implementation, if it's supported by the kernel.
> If it's not supported (< 5.9), use a no-op fallback. Looping over 2^31
> file descriptors calling close() on them is probably not a good idea.
> 
> Reported-by: lemmi <lemmi@nerd2nerd.org>
> Signed-off-by: Stefano Brivio <sbrivio@redhat.com>

Reviewed-by: David Gibson <david@gibson.dropbear.id.au>

> ---
>  util.h | 20 ++++++++++++++++++++
>  1 file changed, 20 insertions(+)
> 
> diff --git a/util.h b/util.h
> index cb4d181..a9deba5 100644
> --- a/util.h
> +++ b/util.h
> @@ -14,6 +14,9 @@
>  #include <string.h>
>  #include <signal.h>
>  #include <arpa/inet.h>
> +#include <unistd.h>
> +#include <sys/syscall.h>
> +#include <linux/close_range.h>
>  
>  #include "log.h"
>  
> @@ -160,6 +163,23 @@ struct ctx;
>  
>  /* cppcheck-suppress funcArgNamesDifferent */
>  __attribute__ ((weak)) int ffsl(long int i) { return __builtin_ffsl(i); }
> +
> +#ifdef CLOSE_RANGE_UNSHARE	/* Linux kernel >= 5.9 */
> +/* glibc < 2.34 and musl as of 1.2.5 need these */
> +#ifndef SYS_close_range
> +#define SYS_close_range		436
> +#endif
> +__attribute__ ((weak)) int close_range(unsigned int first, unsigned int last,
> +				       int flags) {
> +	return syscall(SYS_close_range, first, last, flags);
> +}
> +#else
> +/* No reasonable fallback option */
> +int close_range(unsigned int first, unsigned int last, int flags) {
> +	return 0;
> +}
> +#endif
> +
>  int sock_l4_sa(const struct ctx *c, enum epoll_type type,
>  	       const void *sa, socklen_t sl,
>  	       const char *ifname, bool v6only, uint32_t data);

-- 
David Gibson (he or they)	| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you, not the other way
				| around.
http://www.ozlabs.org/~dgibson