KIOXIA KBG40ZNV256G stuck after resume

[P BS]

Hi,

About a year ago, I filed this bug

    https://bugzilla.kernel.org/show_bug.cgi?id=214983
    KIOXIA KBG40ZNV256G NVME SSD killed by resume from s2idle

about an SSD that wouldn't wake up after suspend. While the bug is
still open, I'd managed to work around it up to now by setting
iommu=soft. But somewhere between 5.18.16 and 5.19.1, that workaround
broke. So following Keith Busch's suggestion in that bug report, I'm
taking the issue here. I'm hoping that given the narrow range of
commits, somebody will be able to identify the commit that likely
caused the regression. Otherwise, I'll bisect it and report the
results here.

Philip