On 04.12.24 11:10, Przemek Kitszel wrote:
On 12/3/24 21:28, Gerhard Engleder wrote:
From: Gerhard Engleder <[email protected]>
From: Gerhard Engleder <[email protected]>
duplicated From: line
Nervous fingers, sorry, will be fixed.
Link down and up triggers update of MTA table. This update executes many
PCIe writes and a final flush. Thus, PCIe will be blocked until all
writes
are flushed. As a result, DMA transfers of other targets suffer from
delay
in the range of 50us. This results in timing violations on real-time
systems during link down and up of e1000e.
A flush after a low enough number of PCIe writes eliminates the delay
but also increases the time needed for MTA table update. The following
measurements were done on i3-2310E with e1000e for 128 MTA table entries:
Single flush after all writes: 106us
Flush after every write: 429us
Flush after every 2nd write: 266us
Flush after every 4th write: 180us
Flush after every 8th write: 141us
Flush after every 16th write: 121us
A flush after every 8th write delays the link up by 35us and the
negative impact to DMA transfers of other targets is still tolerable.
Execute a flush after every 8th write. This prevents overloading the
interconnect with posted writes. As this also increases the time spent
for
MTA table update considerable this change is limited to PREEMPT_RT.
hmm, why to limit this to PREEMPT_RT, the change sounds resonable also
for the standard kernel, at last for me
(perhaps with every 16th write instead)
As Andrew argumented similar, I will remove the PREEMPT_RT dependency
with the next version. This is not the hot path, so the additional delay
of <<1ms for boot and interface up is negligible.
with that said, I'm fine with this patch as is too
Signed-off-by: Gerhard Engleder <[email protected]>
would be good to add link to your RFC
https://lore.kernel.org/netdev/[email protected]/T/
and also CC Vitaly who participated there (done),
same for IWL mailing list (also CCd), and use iwl-next tag for your
future contributions to intel ethernet
Will be done.
Thank you for the review!
Gerhard
