[PATCH] asm-out0: fix a test failure on 32-bit systems

[PATCH] asm-out0: fix a test failure on 32-bit systems

[Ramsay Jones]

Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
---

Hi Luc,

So, I did some 32-bit testing this afternoon ... ;-)

The problem here being s/symaddr.64/symaddr.32/, of course. I did the
minimum necessary to get it to work (adding the -m64), so no creative
solutions here (not that I could think of any)!

[I suppose that I could also have created asm-out1.c as a copy and then
done s/symaddr.64/symaddr.32/, s/-m64/-m32/, but I didn't.]

ATB,
Ramsay Jones

 validation/linear/asm-out0.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/validation/linear/asm-out0.c b/validation/linear/asm-out0.c
index a8e0be69..8b0907b3 100644
--- a/validation/linear/asm-out0.c
+++ b/validation/linear/asm-out0.c
@@ -6,7 +6,7 @@ static void asm_out0(void)
 
 /*
  * check-name: asm-out0
- * check-command: test-linearize -fdump-ir $file
+ * check-command: test-linearize -m64 -fdump-ir $file
  *
  * check-output-start
 asm_out0:
-- 
2.30.0

Re: [PATCH] asm-out0: fix a test failure on 32-bit systems

[Luc Van Oostenryck]

On Sat, Feb 27, 2021 at 05:15:19PM +0000, Ramsay Jones wrote:
> 
> Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
> ---
> 
> Hi Luc,
> 
> So, I did some 32-bit testing this afternoon ... ;-)

Good :)
 
> The problem here being s/symaddr.64/symaddr.32/, of course. I did the
> minimum necessary to get it to work (adding the -m64), so no creative
> solutions here (not that I could think of any)!
> 
> [I suppose that I could also have created asm-out1.c as a copy and then
> done s/symaddr.64/symaddr.32/, s/-m64/-m32/, but I didn't.]

The ideal solution, at least from a functional point of view, would be
to change the "32" by something like "$POINTER_SIZE$" and doing the
substitution when comparing the files but that would complicate a bit
more the testing and make it slower. Another solution would simply to
add for such tests an option to not display the instruction's size.

In this case, what you did is, IMO, the correct solution because the test
is obviously not size dependent, it's just that the size is displayed.

Thanks,
-- Luc