On 06/01/2016 04:39 PM, Markus Armbruster wrote:
Cao jin <caoj.f...@cn.fujitsu.com> writes:


Not covered:

    static void intel_hda_update_irq(IntelHDAState *d)
    {
-->    int msi = d->msi && msi_enabled(&d->pci);
        int level;

        intel_hda_update_int_sts(d);
        if (d->int_sts & (1U << 31) && d->int_ctl & (1U << 31)) {
            level = 1;
        } else {
            level = 0;
        }
        dprint(d, 2, "%s: level %d [%s]\n", __FUNCTION__,
               level, msi ? "msi" : "intx");
        if (msi) {
            if (level) {
                msi_notify(&d->pci, 0);
            }
        } else {
            pci_set_irq(&d->pci, level);
        }
    }

This is wrong, because the meaning of the test changes from

     (user didn't specify msi=off) && (guest enabled MSI)

to

     (user didn't specify msi=on or msi=off) && (guest enabled MSI)


Thanks for pointing the error, seems I lost the fact that ON_OFF_AUTO_AUTO equals 0.

What about:

        int msi = msi_enabled(&d->pci);

If I understand it correctly, it can only be true when we added MSI
capability, and we do that only when msi=auto (default) or msi=on.


I think the modification is ok.

--
Yours Sincerely,

Cao jin



Reply via email to