[LTP] [PATCH 2/2] Add test for ASLRn't bug

Petr Vorel pvorel@suse.cz
Tue Jan 30 12:29:53 CET 2024


Hi Martin,

Very nice work.
Reviewed-by: Petr Vorel <pvorel@suse.cz>

...
> +++ b/runtest/kernel_misc
> @@ -15,3 +15,4 @@ zram01 zram01.sh
>  zram02 zram02.sh
>  zram03 zram03
>  umip_basic_test umip_basic_test
> +aslr01 aslr01

I wonder, don't we want to run also with strict mode?
Do you want to avoid failure:
aslr01.c:241: TFAIL: Large lib base address has less than 28 random bits

> +static void find_large_lib(void)
> +{
> +	size_t hpsize, libsize = 0;
> +
> +	read_shared_libraries(find_large_lib_callback, &libsize);
> +
> +	if (!libsize) {
> +		tst_brk(TCONF,
> +			"No dynamic libraries loaded, please use LD_PRELOAD");
> +	}
> +
> +	hpsize = tst_get_hugepage_size();
> +	tst_res(TINFO, "Largest loaded library: %s (%zu bytes)", lib_path,
> +		libsize);
> +
> +	if (!hpsize) {
> +		tst_res(TCONF, "Hugepage support appears to be missing");
> +	} else if (libsize < hpsize) {
> +		tst_res(TCONF, "The largest dynamic library is smaller than "
> +			"hugepage size");

very nit: I would not split the string (better for grep).

> +		tst_res(TCONF, "Please use LD_PRELOAD to add larger library");
> +	}
> +}

...
> +	if (strict_check) {
> +		if (compat) {
> +			kconf_minbits = aslr_kconfigs[ASLR_COMPAT_MINBITS].val;
> +			minbits_path = "/proc/sys/vm/mmap_rnd_compat_bits";
> +		} else {
> +			kconf_minbits = aslr_kconfigs[ASLR_MINBITS].val;
> +			minbits_path = "/proc/sys/vm/mmap_rnd_bits";
> +		}
> +
> +		/*
> +		 * Reading mmap_rnd_bits usually requires root privileges.
> +		 * Fall back to kernel config values if unprivileged.
> +		 */
+1. So this require either only root or working kconfig.

> +		if (!access(minbits_path, R_OK))
> +			SAFE_FILE_SCANF(minbits_path, "%d", &minbits);
> +		else if (!kconf_minbits)
> +			tst_brk(TBROK, "Cannot determine kernel ASLR min bits");
> +		else if (tst_parse_int(kconf_minbits, &minbits, 1, 64))
> +			tst_brk(TBROK, "Invalid kernel ASLR min bits value");
> +	}
> +
> +	find_large_lib();
> +}

Kind regards,
Petr


More information about the ltp mailing list