On Thu, Sep 06, 2018 at 04:48:01PM +0800, Liuyixian (Eason) wrote: > > > On 2018/9/5 13:41, Leon Romanovsky wrote: > > On Sat, Sep 01, 2018 at 11:00:19AM +0800, liuyixian wrote: > >> It will trigger unneccessary interrupts caused by time out > >> if prints inside aeq handle under some configurations. > >> Thus, move all prints out of aeq handle to work queue. > >> > >> Signed-off-by: liuyixian <liuyixian@xxxxxxxxxx> > >> --- > >> drivers/infiniband/hw/hns/hns_roce_device.h | 1 + > >> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 227 ++++++++++++---------------- > >> 2 files changed, 97 insertions(+), 131 deletions(-) > >> > > > > It is safe to delete those prints and leave only for unsupported events, > > instead of spamming dmesg with your debug information. > > > > Thanks > > Hi Leon, > > Current mechanism is that if an asynchronous event happened, the driver will > notify the event to IB core. But there is no hint to user without driver prints > in my previous tests. So, I think these prints is valuable to notify user some > asynchronous/unexpected events had happened. The catch here is that you are not printing unexpected events, but you are printing all *ERROR events, so in case of error user will be spammed by them. In case your code knows how to deal with such events, users don't ever need to know about such events. In case your doesn't handle it correctly, it is the time to fix it. Thanks > > Maybe I mistake something :) > > Thanks > > > >
Attachment:
signature.asc
Description: PGP signature
