From: David Hildenbrand <david@redhat.com>
To: qemu-devel@nongnu.org
Cc: qemu-s390x@nongnu.org, "Cornelia Huck" <cohuck@redhat.com>,
"Alex Bennée" <alex.bennee@linaro.org>,
"Thomas Huth" <thuth@redhat.com>,
"Richard Henderson" <rth@twiddle.net>
Subject: Re: [PATCH v1] tests/tcg: target/s390x: Test MVC
Date: Fri, 20 Sep 2019 14:03:30 +0200 [thread overview]
Message-ID: <5cbd350c-0fb6-2b6b-e64b-04bd7f7f0b65@redhat.com> (raw)
In-Reply-To: <20190919092829.15393-1-david@redhat.com>
On 19.09.19 11:28, David Hildenbrand wrote:
> Let's add a test that especially verifies that no data will be touched
> in case we cross page boundaries and one page access triggers a fault.
>
> Before the fault-safe handling fixes, the test failes with:
> TEST mvc on s390x
> data modified during a fault
> make[2]: *** [../Makefile.target:116: run-mvc] Error 1
>
> Cc: Alex Bennée <alex.bennee@linaro.org>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
> tests/tcg/s390x/Makefile.target | 1 +
> tests/tcg/s390x/mvc.c | 109 ++++++++++++++++++++++++++++++++
> 2 files changed, 110 insertions(+)
> create mode 100644 tests/tcg/s390x/mvc.c
>
> diff --git a/tests/tcg/s390x/Makefile.target b/tests/tcg/s390x/Makefile.target
> index 6a3bfa8b29..241ef28f61 100644
> --- a/tests/tcg/s390x/Makefile.target
> +++ b/tests/tcg/s390x/Makefile.target
> @@ -7,3 +7,4 @@ TESTS+=exrl-trt
> TESTS+=exrl-trtr
> TESTS+=pack
> TESTS+=mvo
> +TESTS+=mvc
> diff --git a/tests/tcg/s390x/mvc.c b/tests/tcg/s390x/mvc.c
> new file mode 100644
> index 0000000000..aa552d52e5
> --- /dev/null
> +++ b/tests/tcg/s390x/mvc.c
> @@ -0,0 +1,109 @@
> +#include <stdint.h>
> +#include <stdlib.h>
> +#include <stdio.h>
> +#include <string.h>
> +#include <sys/mman.h>
> +#include <signal.h>
> +#include <setjmp.h>
> +
> +jmp_buf jmp_env;
> +
> +static void handle_sigsegv(int sig)
> +{
> + siglongjmp(jmp_env, 1);
> +}
> +
> +#define ALLOC_SIZE (2 * 4096)
> +
> +static inline void mvc_256(const char *dst, const char *src)
> +{
> + asm volatile (
> + " mvc 0(256,%[dst]),0(%[src])\n"
> + :
> + : [dst] "d" (dst),
> + [src] "d" (src)
> + : "memory");
> +}
> +
> +int main(void)
> +{
> + char *src, *dst;
> + int i;
> +
> + /* register the SIGSEGV handler */
> + if (signal(SIGSEGV, handle_sigsegv) == SIG_ERR) {
> + fprintf(stderr, "SIGSEGV not registered\n");
> + return 1;
> + }
> +
> + /* prepare the buffers - two consecutive pages */
> + src = valloc(ALLOC_SIZE);
> + dst = valloc(ALLOC_SIZE);
> + memset(src, 0xff, ALLOC_SIZE);
> + memset(dst, 0x0, ALLOC_SIZE);
> +
> + /* protect the second pages */
> + if (mprotect(src + 4096, 4096, PROT_NONE) ||
> + mprotect(dst + 4096, 4096, PROT_NONE)) {
> + fprintf(stderr, "mprotect failed\n");
> + return 1;
> + }
> +
> + /* fault on second destination page */
> + if (sigsetjmp(jmp_env, 1) == 0) {
> + mvc_256(dst + 4096 - 128, src);
> + fprintf(stderr, "fault not triggered\n");
> + return 1;
> + }
> +
> + /* fault on second source page */
> + if (sigsetjmp(jmp_env, 1) == 0) {
> + mvc_256(dst, src + 4096 - 128);
> + fprintf(stderr, "fault not triggered\n");
> + return 1;
> + }
> +
> + /* fault on second source and second destination page */
> + if (sigsetjmp(jmp_env, 1) == 0) {
> + mvc_256(dst + 4096 - 128, src + 4096 - 128);
> + fprintf(stderr, "fault not triggered\n");
> + return 1;
> + }
> +
> + /* restore permissions */
> + if (mprotect(src + 4096, 4096, PROT_READ | PROT_WRITE) ||
> + mprotect(dst + 4096, 4096, PROT_READ | PROT_WRITE)) {
> + fprintf(stderr, "mprotect failed\n");
> + return 1;
> + }
> +
> + /* no data must be touched during the faults */
> + for (i = 0; i < ALLOC_SIZE; i++) {
> + if (src[i] != 0xff || dst[i]) {
> + fprintf(stderr, "data modified during a fault\n");
> + return 1;
> + }
> + }
> +
> + /* test if MVC works now correctly accross page boundaries */
> + mvc_256(dst + 4096 - 128, src + 4096 - 128);
> + for (i = 0; i < ALLOC_SIZE; i++) {
> + if (src[i] != 0xff) {
> + fprintf(stderr, "src modified\n");
> + return 1;
> + }
> + if (i < 4096 - 128 || i >= 4096 + 128) {
> + if (dst[i]) {
> + fprintf(stderr, "wrong dst modified\n");
> + return 1;
> + }
> + } else {
> + if (dst[i] != 0xff) {
> + fprintf(stderr, "wrong data moved\n");
> + return 1;
> + }
> + }
> + }
> +
> + return 0;
> +}
>
I'll send this along with the updated pull request of
"[PULL SUBSYSTEM s390x 00/29] s390x/tcg: mem_helper: Fault-safe handling"
--
Thanks,
David / dhildenb
next prev parent reply other threads:[~2019-09-20 12:05 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-19 9:28 [Qemu-devel] [PATCH v1] tests/tcg: target/s390x: Test MVC David Hildenbrand
2019-09-19 9:41 ` David Hildenbrand
2019-09-19 16:01 ` Richard Henderson
2019-09-20 9:14 ` David Hildenbrand
2019-09-20 12:03 ` David Hildenbrand [this message]
2019-09-20 13:02 ` Alex Bennée
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5cbd350c-0fb6-2b6b-e64b-04bd7f7f0b65@redhat.com \
--to=david@redhat.com \
--cc=alex.bennee@linaro.org \
--cc=cohuck@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-s390x@nongnu.org \
--cc=rth@twiddle.net \
--cc=thuth@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).