diff mbox

[v2] i.MX: Fix FEC/ENET receive funtions

Message ID 20180113113445.2705-1-jcd@tribudubois.net (mailing list archive)
State New, archived
Headers show

Commit Message

Jean-Christophe Dubois Jan. 13, 2018, 11:34 a.m. UTC 
The actual imx_eth_enable_rx() function is buggy.

It updates s->regs[ENET_RDAR] after calling qemu_flush_queued_packets().

qemu_flush_queued_packets() is going to call imx_XXX_receive() which itself
is going to call imx_eth_enable_rx().

By updating s->regs[ENET_RDAR] after calling qemu_flush_queued_packets()
we end up updating the register with an outdated value which might
lead to disabling the receive function in the i.MX FEC/ENET device.

This patch change the place where the register update is done so that the
register value stays up to date and the receive function can keep
running.

Reported-by: Fyleo <fyleo45@gmail.com>
Tested-by: Fyleo  <fyleo45@gmail.com>
Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net>
---
Change since v1:
1. Rebase the patch on the updated master branch

 hw/net/imx_fec.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

Comments

Peter Maydell Jan. 22, 2018, 11:48 a.m. UTC | #1 
On 13 January 2018 at 11:34, Jean-Christophe Dubois <jcd@tribudubois.net> wrote:
> The actual imx_eth_enable_rx() function is buggy.
>
> It updates s->regs[ENET_RDAR] after calling qemu_flush_queued_packets().
>
> qemu_flush_queued_packets() is going to call imx_XXX_receive() which itself
> is going to call imx_eth_enable_rx().
>
> By updating s->regs[ENET_RDAR] after calling qemu_flush_queued_packets()
> we end up updating the register with an outdated value which might
> lead to disabling the receive function in the i.MX FEC/ENET device.
>
> This patch change the place where the register update is done so that the
> register value stays up to date and the receive function can keep
> running.
>
> Reported-by: Fyleo <fyleo45@gmail.com>
> Tested-by: Fyleo  <fyleo45@gmail.com>
> Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net>
> ---

Andrey, do you have an opinion on this patch, since you've been
looking at i.MX code recently?

> Change since v1:
> 1. Rebase the patch on the updated master branch
>
>  hw/net/imx_fec.c | 8 ++------
>  1 file changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/hw/net/imx_fec.c b/hw/net/imx_fec.c
> index 4fb48f62ba..9506f9b69f 100644
> --- a/hw/net/imx_fec.c
> +++ b/hw/net/imx_fec.c
> @@ -595,19 +595,16 @@ static void imx_eth_do_tx(IMXFECState *s, uint32_t index)
>  static void imx_eth_enable_rx(IMXFECState *s, bool flush)
>  {
>      IMXFECBufDesc bd;
> -    bool rx_ring_full;
>
>      imx_fec_read_bd(&bd, s->rx_descriptor);
>
> -    rx_ring_full = !(bd.flags & ENET_BD_E);
> +    s->regs[ENET_RDAR] = (bd.flags & ENET_BD_E) ? ENET_RDAR_RDAR : 0;
>
> -    if (rx_ring_full) {
> +    if (!s->regs[ENET_RDAR]) {
>          FEC_PRINTF("RX buffer full\n");
>      } else if (flush) {
>          qemu_flush_queued_packets(qemu_get_queue(s->nic));
>      }
> -
> -    s->regs[ENET_RDAR] = rx_ring_full ? 0 : ENET_RDAR_RDAR;
>  }
>
>  static void imx_eth_reset(DeviceState *d)
> @@ -866,7 +863,6 @@ static void imx_eth_write(void *opaque, hwaddr offset, uint64_t value,
>      case ENET_RDAR:
>          if (s->regs[ENET_ECR] & ENET_ECR_ETHEREN) {
>              if (!s->regs[index]) {
> -                s->regs[index] = ENET_RDAR_RDAR;
>                  imx_eth_enable_rx(s, true);
>              }
>          } else {
> --
> 2.14.1

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM
Andrey Smirnov Jan. 22, 2018, 2:54 p.m. UTC | #2 
On Mon, Jan 22, 2018 at 3:48 AM, Peter Maydell <peter.maydell@linaro.org> wrote:
> On 13 January 2018 at 11:34, Jean-Christophe Dubois <jcd@tribudubois.net> wrote:
>> The actual imx_eth_enable_rx() function is buggy.
>>
>> It updates s->regs[ENET_RDAR] after calling qemu_flush_queued_packets().
>>
>> qemu_flush_queued_packets() is going to call imx_XXX_receive() which itself
>> is going to call imx_eth_enable_rx().
>>
>> By updating s->regs[ENET_RDAR] after calling qemu_flush_queued_packets()
>> we end up updating the register with an outdated value which might
>> lead to disabling the receive function in the i.MX FEC/ENET device.
>>
>> This patch change the place where the register update is done so that the
>> register value stays up to date and the receive function can keep
>> running.
>>
>> Reported-by: Fyleo <fyleo45@gmail.com>
>> Tested-by: Fyleo  <fyleo45@gmail.com>
>> Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net>
>> ---
>
> Andrey, do you have an opinion on this patch, since you've been
> looking at i.MX code recently?
>

The rationale makes sense to me and patch looks like a good cleanup in
general, so FWIW:

Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>

I also gave it a spin against my i.MX7 changes with doing basic things
like ping and scp of 1GB file, so I can give my:

Tested-by: Andrey Smirnov <andrew.smirnov@gmail.com>

for that.

Thanks,
Andrey Smirnov


>> Change since v1:
>> 1. Rebase the patch on the updated master branch
>>
>>  hw/net/imx_fec.c | 8 ++------
>>  1 file changed, 2 insertions(+), 6 deletions(-)
>>
>> diff --git a/hw/net/imx_fec.c b/hw/net/imx_fec.c
>> index 4fb48f62ba..9506f9b69f 100644
>> --- a/hw/net/imx_fec.c
>> +++ b/hw/net/imx_fec.c
>> @@ -595,19 +595,16 @@ static void imx_eth_do_tx(IMXFECState *s, uint32_t index)
>>  static void imx_eth_enable_rx(IMXFECState *s, bool flush)
>>  {
>>      IMXFECBufDesc bd;
>> -    bool rx_ring_full;
>>
>>      imx_fec_read_bd(&bd, s->rx_descriptor);
>>
>> -    rx_ring_full = !(bd.flags & ENET_BD_E);
>> +    s->regs[ENET_RDAR] = (bd.flags & ENET_BD_E) ? ENET_RDAR_RDAR : 0;
>>
>> -    if (rx_ring_full) {
>> +    if (!s->regs[ENET_RDAR]) {
>>          FEC_PRINTF("RX buffer full\n");
>>      } else if (flush) {
>>          qemu_flush_queued_packets(qemu_get_queue(s->nic));
>>      }
>> -
>> -    s->regs[ENET_RDAR] = rx_ring_full ? 0 : ENET_RDAR_RDAR;
>>  }
>>
>>  static void imx_eth_reset(DeviceState *d)
>> @@ -866,7 +863,6 @@ static void imx_eth_write(void *opaque, hwaddr offset, uint64_t value,
>>      case ENET_RDAR:
>>          if (s->regs[ENET_ECR] & ENET_ECR_ETHEREN) {
>>              if (!s->regs[index]) {
>> -                s->regs[index] = ENET_RDAR_RDAR;
>>                  imx_eth_enable_rx(s, true);
>>              }
>>          } else {
>> --
>> 2.14.1
>
> Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
>
> thanks
> -- PMM
Peter Maydell Jan. 22, 2018, 2:59 p.m. UTC | #3 
On 22 January 2018 at 14:54, Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
> On Mon, Jan 22, 2018 at 3:48 AM, Peter Maydell <peter.maydell@linaro.org> wrote:
>> On 13 January 2018 at 11:34, Jean-Christophe Dubois <jcd@tribudubois.net> wrote:
>>> The actual imx_eth_enable_rx() function is buggy.
>>>
>>> It updates s->regs[ENET_RDAR] after calling qemu_flush_queued_packets().
>>>
>>> qemu_flush_queued_packets() is going to call imx_XXX_receive() which itself
>>> is going to call imx_eth_enable_rx().
>>>
>>> By updating s->regs[ENET_RDAR] after calling qemu_flush_queued_packets()
>>> we end up updating the register with an outdated value which might
>>> lead to disabling the receive function in the i.MX FEC/ENET device.
>>>
>>> This patch change the place where the register update is done so that the
>>> register value stays up to date and the receive function can keep
>>> running.
>>>
>>> Reported-by: Fyleo <fyleo45@gmail.com>
>>> Tested-by: Fyleo  <fyleo45@gmail.com>
>>> Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net>
>>> ---
>>
>> Andrey, do you have an opinion on this patch, since you've been
>> looking at i.MX code recently?
>>
>
> The rationale makes sense to me and patch looks like a good cleanup in
> general, so FWIW:
>
> Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>
>
> I also gave it a spin against my i.MX7 changes with doing basic things
> like ping and scp of 1GB file, so I can give my:
>
> Tested-by: Andrey Smirnov <andrew.smirnov@gmail.com>

Thanks; I've applied the patch to target-arm.next.

-- PMM
Jean-Christophe Dubois Jan. 24, 2018, 8:07 p.m. UTC | #4 
My guess is that with this patch, the "flush" feature that was added by 
Andrey in "imx_fec: Change queue flushing heuristics" (commit b2b012a) 
is not really necessary anymore.

But it does not hurt (it might induce a little bit more processing).

JC

Le 22/01/2018 à 15:59, Peter Maydell a écrit :
> On 22 January 2018 at 14:54, Andrey Smirnov <andrew.smirnov@gmail.com> wrote:
>> On Mon, Jan 22, 2018 at 3:48 AM, Peter Maydell <peter.maydell@linaro.org> wrote:
>>> On 13 January 2018 at 11:34, Jean-Christophe Dubois <jcd@tribudubois.net> wrote:
>>>> The actual imx_eth_enable_rx() function is buggy.
>>>>
>>>> It updates s->regs[ENET_RDAR] after calling qemu_flush_queued_packets().
>>>>
>>>> qemu_flush_queued_packets() is going to call imx_XXX_receive() which itself
>>>> is going to call imx_eth_enable_rx().
>>>>
>>>> By updating s->regs[ENET_RDAR] after calling qemu_flush_queued_packets()
>>>> we end up updating the register with an outdated value which might
>>>> lead to disabling the receive function in the i.MX FEC/ENET device.
>>>>
>>>> This patch change the place where the register update is done so that the
>>>> register value stays up to date and the receive function can keep
>>>> running.
>>>>
>>>> Reported-by: Fyleo <fyleo45@gmail.com>
>>>> Tested-by: Fyleo  <fyleo45@gmail.com>
>>>> Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net>
>>>> ---
>>> Andrey, do you have an opinion on this patch, since you've been
>>> looking at i.MX code recently?
>>>
>> The rationale makes sense to me and patch looks like a good cleanup in
>> general, so FWIW:
>>
>> Reviewed-by: Andrey Smirnov <andrew.smirnov@gmail.com>
>>
>> I also gave it a spin against my i.MX7 changes with doing basic things
>> like ping and scp of 1GB file, so I can give my:
>>
>> Tested-by: Andrey Smirnov <andrew.smirnov@gmail.com>
> Thanks; I've applied the patch to target-arm.next.
>
> -- PMM
>
>
diff mbox

Patch

diff --git a/hw/net/imx_fec.c b/hw/net/imx_fec.c
index 4fb48f62ba..9506f9b69f 100644
--- a/hw/net/imx_fec.c
+++ b/hw/net/imx_fec.c
@@ -595,19 +595,16 @@  static void imx_eth_do_tx(IMXFECState *s, uint32_t index)
 static void imx_eth_enable_rx(IMXFECState *s, bool flush)
 {
     IMXFECBufDesc bd;
-    bool rx_ring_full;
 
     imx_fec_read_bd(&bd, s->rx_descriptor);
 
-    rx_ring_full = !(bd.flags & ENET_BD_E);
+    s->regs[ENET_RDAR] = (bd.flags & ENET_BD_E) ? ENET_RDAR_RDAR : 0;
 
-    if (rx_ring_full) {
+    if (!s->regs[ENET_RDAR]) {
         FEC_PRINTF("RX buffer full\n");
     } else if (flush) {
         qemu_flush_queued_packets(qemu_get_queue(s->nic));
     }
-
-    s->regs[ENET_RDAR] = rx_ring_full ? 0 : ENET_RDAR_RDAR;
 }
 
 static void imx_eth_reset(DeviceState *d)
@@ -866,7 +863,6 @@  static void imx_eth_write(void *opaque, hwaddr offset, uint64_t value,
     case ENET_RDAR:
         if (s->regs[ENET_ECR] & ENET_ECR_ETHEREN) {
             if (!s->regs[index]) {
-                s->regs[index] = ENET_RDAR_RDAR;
                 imx_eth_enable_rx(s, true);
             }
         } else {