[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 04/17] imx_fec: Change queue flushing heuristics
From: |
Andrey Smirnov |
Subject: |
Re: [Qemu-devel] [PATCH 04/17] imx_fec: Change queue flushing heuristics |
Date: |
Mon, 25 Sep 2017 11:10:04 -0700 |
On Fri, Sep 22, 2017 at 12:27 AM, Jason Wang <address@hidden> wrote:
>
>
> On 2017年09月19日 03:50, Andrey Smirnov wrote:
>>
>> In current implementation, packet queue flushing logic seem to suffer
>> from a deadlock like scenario if a packet is received by the interface
>> before before Rx ring is initialized by Guest's driver. Consider the
>> following sequence of events:
>>
>> 1. A QEMU instance is started against a TAP device on Linux
>> host, running Linux guest, e. g., something to the effect
>> of:
>>
>> qemu-system-arm \
>> -net nic,model=imx.fec,netdev=lan0 \
>> netdev tap,id=lan0,ifname=tap0,script=no,downscript=no \
>> ... rest of the arguments ...
>>
>> 2. Once QEMU starts, but before guest reaches the point where
>> FEC deriver is done initializing the HW, Guest, via TAP
>> interface, receives a number of multicast MDNS packets from
>> Host (not necessarily true for every OS, but it happens at
>> least on Fedora 25)
>>
>> 3. Recieving a packet in such a state results in
>> imx_eth_can_receive() returning '0', which in turn causes
>> tap_send() to disable corresponding event (tap.c:203)
>>
>> 4. Once Guest's driver reaches the point where it is ready to
>> recieve packets it prepares Rx ring descriptors and writes
>> ENET_RDAR_RDAR to ENET_RDAR register to indicate to HW that
>> more descriptors are ready. And at this points emulation
>> layer does this:
>>
>> s->regs[index] = ENET_RDAR_RDAR;
>> imx_eth_enable_rx(s);
>>
>> which, combined with:
>>
>> if (!s->regs[ENET_RDAR]) {
>> qemu_flush_queued_packets(qemu_get_queue(s->nic));
>> }
>
>
> Not familiar with FEC, but if you are tracking 0->1 transition, why not
> simply introduce a parameter of imx_eth_enable_rx() to force the flushing?
>
Not sure I fully understand you, are you proposing I get rid of
"needs_flush" parameter in the device state, converting it to be a
parameter to imx_eth_enable_rx(), and then force flushing every time
imx_eth_enable_rx() is called in imx_eth_write()?
That should work, but it'll end up making the emulator code to flush
corresponding NIC queue every time the driver is done processing RX
ring. If that is not a big problem I am more than happy to make that
change.
Thanks,
Andrey Smirnov
- [Qemu-devel] [PATCH 00/17] Initial i.MX7 support, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 03/17] imx_fec: Refactor imx_eth_enable_rx(), Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 01/17] imx_fec: Do not link to netdev, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 04/17] imx_fec: Change queue flushing heuristics, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 02/17] imx_fec: Do not calculate FEC, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 06/17] imx_fec: Use MIN instead of explicit ternary operator, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 05/17] imx_fec: Use ENET_FTRL to determine truncation length, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 09/17] imx_fec: Use correct length for packet size, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 08/17] imx_fec: Add support for multiple Tx DMA rings, Andrey Smirnov, 2017/09/18
- [Qemu-devel] [PATCH 07/17] imx_fec: Emulate SHIFT16 in ENETx_RACC, Andrey Smirnov, 2017/09/18