On 06/09/2022 18.31, Patrick Venture wrote:
> The register tests walks all the registers to verify they are initially
> 0 when appropriate. However, if the MAC address is set in the register
> space, this should not be checked against 0.
>
> Reviewed-by: Hao Wu <wuhaotsh@google.com>
> Change-Id: I02426e39bdab33ceedd42c49d233e8680d4ec058
What's that change-id good for?
Oops, sorry about that. I can send out a v2 without it, or during application someone can nicely trim it? :)
> Signed-off-by: Patrick Venture <venture@google.com>
> ---
> tests/qtest/npcm7xx_emc-test.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/tests/qtest/npcm7xx_emc-test.c b/tests/qtest/npcm7xx_emc-test.c
> index 7c435ac915..207d8515b7 100644
> --- a/tests/qtest/npcm7xx_emc-test.c
> +++ b/tests/qtest/npcm7xx_emc-test.c
> @@ -378,7 +378,8 @@ static void test_init(gconstpointer test_data)
>
> #undef CHECK_REG
>
> - for (i = 0; i < NUM_CAMML_REGS; ++i) {
> + /* Skip over the MAC address registers, which is BASE+0 */
> + for (i = 1; i < NUM_CAMML_REGS; ++i) {
> g_assert_cmpuint(emc_read(qts, mod, REG_CAMM_BASE + i * 2), ==,
> 0);
> g_assert_cmpuint(emc_read(qts, mod, REG_CAML_BASE + i * 2), ==,
Basically ack, but one question: Where should that non-zero MAC address come
from / when did you hit a problem here? If QEMU is started without any mac
settings at all (like it is done here), the register never contains a
non-zero value, does it?
So, there's a bug in the emc device presently where that value isn't set when it should be. I have that bug fixed, but for whatever reason, probably not enough caffeine, I didn't bundle the two patches together.
Thomas