diff options
author | David S. Miller <davem@davemloft.net> | 2020-09-18 17:47:07 -0700 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2020-09-18 17:47:07 -0700 |
commit | 5e43df14d6b436ca767676bc8e41506dc03c4d89 (patch) | |
tree | ad54a1a0640cf8a23c647f75989e9012d177eeb5 /tools/perf/scripts/python/export-to-postgresql.py | |
parent | ba4ee3c053659119472135231dbef8f6880ce1fb (diff) | |
parent | 54f7e44353a7319a8f779fdc1ed9ec02c5eaf826 (diff) |
Merge branch 'net-ipa-wake-up-system-on-RX-available'
Alex Elder says:
====================
net: ipa: wake up system on RX available
This series arranges for the IPA driver to wake up a suspended
system if the IPA hardware has a packet to deliver to the AP.
Version 2 replaced the first patch from version 1 with three
patches, in response to David Miller's feedback. And based on
Bjorn Andersson's feedback on version 2, this version reworks
the tracking of IPA clock references. As a result, we no
longer need a flag to determine whether a "don't' suspend" clock
reference is held (though an bit in a bitmask is still used for
a different purpose).
In summary:
- A refcount_t is used to track IPA clock references where an
atomic_t was previously used. (This may go away soon as well,
with upcoming work to implement runtime PM.)
- We no longer track whether a special reference has been taken
to avoid suspending IPA.
- A bit in a bitmask is used to ensure we only trigger a system
resume once per system suspend.
And from the original series:
- Suspending endpoints only occurs when suspending the driver,
not when dropping the last clock reference. Resuming
endpoints is also disconnected from starting the clock.
- The IPA SUSPEND interrupt is now a wakeup interrupt. If it
fires, it schedules a system resume operation.
- The GSI interrupt is no longer a wakeup interrupt.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'tools/perf/scripts/python/export-to-postgresql.py')
0 files changed, 0 insertions, 0 deletions