[PATCH] eal: support including mapped memory in core dump

[Stephen Hemminger]

On Thu, 24 Oct 2024 02:18:59 +0300
Dmitry Kozlyuk <dmitry.kozliuk at gmail.com> wrote:

> +static int
> +enable_shared_hugepage_coredump(void)
> +{
> +	const char *path = "/proc/self/coredump_filter";
> +	const unsigned long shared_hugepage_flag = RTE_BIT64(6);
> +
> +	FILE *f;
> +	uint64_t coredump_filter;
> +
> +	f = fopen(path, "r");
> +	if (f == NULL) {
> +		rte_errno = errno;
> +		EAL_LOG(ERR, "Failed to open %s for reading: %s", path, strerror(errno));
> +		return -1;
> +	}
> +
> +	if (fscanf(f, "%"SCNx64, &coredump_filter) != 1) {
> +		rte_errno = errno;
> +		EAL_LOG(ERR, "Failed to parse %s: %s", path, strerror(errno));
> +		fclose(f);
> +		return -1;
> +	}
> +
> +	fclose(f);
> +
> +	if (coredump_filter & shared_hugepage_flag)
> +		return 0;
> +
> +	f = fopen(path, "w");
> +	if (f == NULL) {
> +		rte_errno = errno;
> +		EAL_LOG(ERR, "Failed to open %s for writing: %s", path, strerror(errno));
> +		return -1;
> +	}
> +
> +	coredump_filter |= shared_hugepage_flag;
> +	if (fprintf(f, "%"PRIx64, coredump_filter) <= 0) {
> +		rte_errno = EIO;
> +		EAL_LOG(ERR, "Failed to write %"PRIx64" to %s", coredump_filter, path);
> +		fclose(f);
> +		return -1;
> +	}
> +
> +	fclose(f);
> +	return 0;
> +}
> +

Having a process set a system global value like coredump_filter via an internal
call seems like a potential problem. What about other processes on the system?
It may not even be allowed if using a hardened kernel.

I would prefer that madvise() be used, and document the required change to
coredump_filter.