MSI interrupts are messages to the APIC, so they are more like edge
interrupts, rather than normal level PCI interrupts. The interrupt cause
in the e1000 driver will be cleared as soon as the interrupt is serviced
by the driver.



On 2016-04-12, 11:35 AM, "Sahoo, Manas R (GE, Intelligent Platforms)"
<community-noreply@qnx.com> wrote:

>Ok I was thinking that. But then I was not clear about enabling
>interrupts in MSI case. The FPGA & e1000, both are using MSIs. When the
>MSI will get cleared? Once you enter the ISR, is that time it gets
>cleared.
>
>-----Original Message-----
>From: Hugh Brown [mailto:community-noreply@qnx.com]
>Sent: Tuesday, April 12, 2016 8:42 PM
>To: drivers-networking@community.qnx.com
>Subject: Re: e1000 driver ISR timings
>
>Have you tried starting your FPGA driver before io-pkt? I could be due to
>an interrupt priority issue.
>
>
>
>On 2016-04-12, 11:01 AM, "Sahoo, Manas R (GE, Intelligent Platforms)"
><community-noreply@qnx.com> wrote:
>
>>Does that mean I need to start e1000 driver with specific options, so
>>that I will see a constant ISR service time. And also how come my FRC
>>ISR is getting preempted by e1000 ISR.
>>
>>-----Original Message-----
>>From: Hugh Brown [mailto:community-noreply@qnx.com]
>>Sent: Tuesday, April 12, 2016 8:20 PM
>>To: drivers-networking@community.qnx.com
>>Subject: Re: e1000 driver ISR timings
>>
>>The e1000 interrupt service routine is only used for receive or link
>>events. The ISR time will depend on whether there is just one buffer to
>>receive or multiple buffers.
>>
>>
>>
>>On 2016-04-12, 9:02 AM, "manas sahoo" <community-noreply@qnx.com> wrote:
>>
>>>Hi,
>>>In my system, i have e1000 driver and another timer interrrupt (due to
>>>FPGA FRC). The e1000 has interrupt number of 0x104 and the FRC is 0x10a.
>>>While running the system, we are seeing that sometimes the FRC ISR is
>>>preempted by e1000 and after e1000 ISR runs, then FRC ISR gets time.
>>>Also, at sometimes the e1000 ISR takes arround 40us to execute and
>>>most of the times around 2us to complete.
>>>
>>>Is this expected behavior or are there any concerns around this with
>>>e1000?
>>>
>>>Thanks,
>>>Manas
>>>
>>>
>>>
>>>_______________________________________________
>>>
>>>Networking Drivers
>>>https://urldefense.proofpoint.com/v2/url?u=http-3A__community.qnx.com_
>>>s 
>>>f_go_post116080&d=CwIFAg&c=IV_clAzoPDE253xZdHuilRgztyh_RiV3wUrLrDQYWSI
>>>& 
>>>r=x4M_FqHWSd9lCjYx6OSUWUXZesy3o-swjtVkHTWouHQ&m=qxKTS7sR-5P8STVJyGHqqT
>>>g 
>>>wizqtxeT0OBcB-qiQNZM&s=Cbn4igG41hy201Ejbm8T7eEDWNrEM0JQxcKavsfQfVg&e=
>>>To cancel your subscription to this discussion, please e-mail
>>>drivers-networking-unsubscribe@community.qnx.com
>>
>>
>>
>>
>>
>>_______________________________________________
>>
>>Networking Drivers
>>https://urldefense.proofpoint.com/v2/url?u=http-3A__community.qnx.com_s
>>f_g 
>>o_post116081&d=CwIFAg&c=IV_clAzoPDE253xZdHuilRgztyh_RiV3wUrLrDQYWSI&r=x
>>4M_
>>FqHWSd9lCjYx6OSUWUXZesy3o-swjtVkHTWouHQ&m=qxKTS7sR-5P8STVJyGHqqTgwizqtx
>>eT0 OBcB-qiQNZM&s=I9-_nIqoK1f6AQERwaX7olequjIyoT4rMzZNzv8UI48&e=
>>To cancel your subscription to this discussion, please e-
mail
>>drivers-networking-unsubscribe@community.qnx.com
>>
>>
>>
>>
>>_______________________________________________
>>
>>Networking...