Skip to contentSkip to conversation

net: smc91c111 can_receive fixes

PATCH Applied Peter Crosthwaite 4 participants 11 messages Sep 11, 2015 – Sep 17, 2015

lore
Share & tools
lore
#0 lore raw

cover net: smc91c111 can_receive fixes

Hi Richard,

This should hopefully fix your bug, while addressing the extra concern I raised.

There was also inconsistent behaviour with corking packets through a soft reset which I notice and fixed.

Please let me know if this works for you.

Regards,
Peter

Peter Crosthwaite (3):
net: smc91c111: guard flush_queued_packets() on can_rx()
net: smc91c111: gate can_receive() on rx FIFO having a slot
net: smc91c111: flush packets on RCR register changes

 hw/net/smc91c111.c | 33 +++++++++++++++++++++++++--------
 1 file changed, 25 insertions(+), 8 deletions(-)
Signature · 1 line
1.9.1
#1 lore raw

PATCH 1/3 net: smc91c111: guard flush_queued_packets() on can_rx()

Check that the core can once again receive packets before asking the net layer to do a flush. This will make it more convenient to flush packets when adding new conditions to can_receive.

Add missing if braces while moving the can_receive() core code.

---

 hw/net/smc91c111.c | 30 ++++++++++++++++++++++--------
 1 file changed, 22 insertions(+), 8 deletions(-)
hw/net/smc91c111.c +22 -8
index 74e06e6..5774eff 100644
@@ -124,6 +124,24 @@ static void smc91c111_update(smc91c111_state *s)
     qemu_set_irq(s->irq, level);
 }
 
+static int smc91c111_can_receive(smc91c111_state *s)
+{
+    if ((s->rcr & RCR_RXEN) == 0 || (s->rcr & RCR_SOFT_RST)) {
+        return 1;
+    }
+    if (s->allocated == (1 << NUM_PACKETS) - 1) {
+        return 0;
+    }
+    return 1;
+}
+
+static inline void smc91c111_flush_queued_packets(smc91c111_state *s)
+{
+    if (smc91c111_can_receive(s)) {
+        qemu_flush_queued_packets(qemu_get_queue(s->nic));
+    }
+}
+
 /* Try to allocate a packet.  Returns 0x80 on failure.  */
 static int smc91c111_allocate_packet(smc91c111_state *s)
 {
@@ -185,7 +203,7 @@ static void smc91c111_release_packet(smc91c111_state *s, int packet)
     s->allocated &= ~(1 << packet);
     if (s->tx_alloc == 0x80)
         smc91c111_tx_alloc(s);
-    qemu_flush_queued_packets(qemu_get_queue(s->nic));
+    smc91c111_flush_queued_packets(s);
 }
 
 /* Flush the TX FIFO.  */
@@ -636,15 +654,11 @@ static uint32_t smc91c111_readl(void *opaque, hwaddr offset)
     return val;
 }
 
-static int smc91c111_can_receive(NetClientState *nc)
+static int smc91c111_can_receive_nc(NetClientState *nc)
 {
     smc91c111_state *s = qemu_get_nic_opaque(nc);
 
-    if ((s->rcr & RCR_RXEN) == 0 || (s->rcr & RCR_SOFT_RST))
-        return 1;
-    if (s->allocated == (1 << NUM_PACKETS) - 1)
-        return 0;
-    return 1;
+    return smc91c111_can_receive(s);
 }
 
 static ssize_t smc91c111_receive(NetClientState *nc, const uint8_t *buf, size_t size)
@@ -739,7 +753,7 @@ static const MemoryRegionOps smc91c111_mem_ops = {
 static NetClientInfo net_smc91c111_info = {
     .type = NET_CLIENT_OPTIONS_KIND_NIC,
     .size = sizeof(NICState),
-    .can_receive = smc91c111_can_receive,
+    .can_receive = smc91c111_can_receive_nc,
     .receive = smc91c111_receive,
 };
 
Signature · 1 line
1.9.1
#3 lore raw

PATCH 2/3 net: smc91c111: gate can_receive() on rx FIFO having a slot

In reply to Peter Crosthwaite

Return false from can_receive() when the FIFO doesn't have a free RX slot. This fixes a bug in the current code where the allocated buffer is freed before the fifo pop, triggering a premature flush of queued RX packets. It also will handle a corner case, where the guest manually frees the allocated buffer before popping the rx FIFO (hence it is not enough to just delay the flush_queued_packets()).

---

 hw/net/smc91c111.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/net/smc91c111.c +3 -1
index 5774eff..8fc3deb 100644
@@ -129,7 +129,8 @@ static int smc91c111_can_receive(smc91c111_state *s)
     if ((s->rcr & RCR_RXEN) == 0 || (s->rcr & RCR_SOFT_RST)) {
         return 1;
     }
-    if (s->allocated == (1 << NUM_PACKETS) - 1) {
+    if (s->allocated == (1 << NUM_PACKETS) - 1 ||
+        s->rx_fifo_len == NUM_PACKETS) {
         return 0;
     }
     return 1;
@@ -182,6 +183,7 @@ static void smc91c111_pop_rx_fifo(smc91c111_state *s)
     } else {
         s->int_level &= ~INT_RCV;
     }
+    smc91c111_flush_queued_packets(s);
     smc91c111_update(s);
 }
 
Signature · 1 line
1.9.1
#5 lore raw

PATCH 3/3 net: smc91c111: flush packets on RCR register changes

In reply to Peter Crosthwaite

The SOFT_RST or RXEN in the control register can be used as a condition to unblock the net layer via can_receive(). So check for possible flushes on RCR changes. This will drop all pending packets on soft reset or disable which is the functional intent of the can_receive() logic.

---

 hw/net/smc91c111.c | 1 +
 1 file changed, 1 insertion(+)
hw/net/smc91c111.c +1 -0
index 8fc3deb..c19cdd1 100644
@@ -331,6 +331,7 @@ static void smc91c111_writeb(void *opaque, hwaddr offset,
             if (s->rcr & RCR_SOFT_RST) {
                 smc91c111_reset(DEVICE(s));
             }
+            smc91c111_flush_queued_packets(s);
             return;
         case 10: case 11: /* RPCR */
             /* Ignored */
Signature · 1 line
1.9.1
#7 lore raw
In reply to Peter Crosthwaite

Hi Peter,

Quoted 7 lines · Peter Crosthwaite

This should hopefully fix your bug, while addressing the extra concern I raised.

There was also inconsistent behaviour with corking packets through a soft reset which I notice and fixed.

Please let me know if this works for you.

I've run it through a few builds/tests in place of my own patches and so far it seems to be working, thanks!

Cheers,

Richard

Reviewed-by on v1 1/3
#2 lore raw

Re: [Qemu-devel] [RFT PATCH v1 1/3] net: smc91c111: guard flush_queued_packets() on can_rx()

In reply to Peter Crosthwaite
Quoted 81 lines · Peter Crosthwaite

Check that the core can once again receive packets before asking the net layer to do a flush. This will make it more convenient to flush packets when adding new conditions to can_receive.

Add missing if braces while moving the can_receive() core code.

---

 hw/net/smc91c111.c | 30 ++++++++++++++++++++++--------
 1 file changed, 22 insertions(+), 8 deletions(-)
hw/net/smc91c111.c +22 -8
index 74e06e6..5774eff 100644
@@ -124,6 +124,24 @@ static void smc91c111_update(smc91c111_state *s)
     qemu_set_irq(s->irq, level);
 }
 
+static int smc91c111_can_receive(smc91c111_state *s)
+{
+    if ((s->rcr & RCR_RXEN) == 0 || (s->rcr & RCR_SOFT_RST)) {
+        return 1;
+    }
+    if (s->allocated == (1 << NUM_PACKETS) - 1) {
+        return 0;
+    }
+    return 1;
+}
+
+static inline void smc91c111_flush_queued_packets(smc91c111_state *s)
+{
+    if (smc91c111_can_receive(s)) {
+        qemu_flush_queued_packets(qemu_get_queue(s->nic));
+    }
+}
+
 /* Try to allocate a packet.  Returns 0x80 on failure.  */
 static int smc91c111_allocate_packet(smc91c111_state *s)
 {
@@ -185,7 +203,7 @@ static void smc91c111_release_packet(smc91c111_state *s, int packet)
     s->allocated &= ~(1 << packet);
     if (s->tx_alloc == 0x80)
         smc91c111_tx_alloc(s);
-    qemu_flush_queued_packets(qemu_get_queue(s->nic));
+    smc91c111_flush_queued_packets(s);
 }
 
 /* Flush the TX FIFO.  */
@@ -636,15 +654,11 @@ static uint32_t smc91c111_readl(void *opaque, hwaddr offset)
     return val;
 }
 
-static int smc91c111_can_receive(NetClientState *nc)
+static int smc91c111_can_receive_nc(NetClientState *nc)
 {
     smc91c111_state *s = qemu_get_nic_opaque(nc);
 
-    if ((s->rcr & RCR_RXEN) == 0 || (s->rcr & RCR_SOFT_RST))
-        return 1;
-    if (s->allocated == (1 << NUM_PACKETS) - 1)
-        return 0;
-    return 1;
+    return smc91c111_can_receive(s);
 }
 
 static ssize_t smc91c111_receive(NetClientState *nc, const uint8_t *buf, size_t size)
@@ -739,7 +753,7 @@ static const MemoryRegionOps smc91c111_mem_ops = {
 static NetClientInfo net_smc91c111_info = {
     .type = NET_CLIENT_OPTIONS_KIND_NIC,
     .size = sizeof(NICState),
-    .can_receive = smc91c111_can_receive,
+    .can_receive = smc91c111_can_receive_nc,
     .receive = smc91c111_receive,
 };
 
Signature · 1 line
1.9.1
Reviewed-by on v1 2/3
#4 lore raw

Re: [Qemu-devel] [RFT PATCH v1 2/3] net: smc91c111: gate can_receive() on rx FIFO having a slot

In reply to Peter Crosthwaite
Quoted 40 lines · Peter Crosthwaite

Return false from can_receive() when the FIFO doesn't have a free RX slot. This fixes a bug in the current code where the allocated buffer is freed before the fifo pop, triggering a premature flush of queued RX packets. It also will handle a corner case, where the guest manually frees the allocated buffer before popping the rx FIFO (hence it is not enough to just delay the flush_queued_packets()).

---

 hw/net/smc91c111.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/net/smc91c111.c +3 -1
index 5774eff..8fc3deb 100644
@@ -129,7 +129,8 @@ static int smc91c111_can_receive(smc91c111_state *s)
     if ((s->rcr & RCR_RXEN) == 0 || (s->rcr & RCR_SOFT_RST)) {
         return 1;
     }
-    if (s->allocated == (1 << NUM_PACKETS) - 1) {
+    if (s->allocated == (1 << NUM_PACKETS) - 1 ||
+        s->rx_fifo_len == NUM_PACKETS) {
         return 0;
     }
     return 1;
@@ -182,6 +183,7 @@ static void smc91c111_pop_rx_fifo(smc91c111_state *s)
     } else {
         s->int_level &= ~INT_RCV;
     }
+    smc91c111_flush_queued_packets(s);
     smc91c111_update(s);
 }
 
Signature · 1 line
1.9.1
Reviewed-by on v1 3/3
#6 lore raw

Re: [Qemu-devel] [RFT PATCH v1 3/3] net: smc91c111: flush packets on RCR register changes

In reply to Peter Crosthwaite
Quoted 28 lines · Peter Crosthwaite

The SOFT_RST or RXEN in the control register can be used as a condition to unblock the net layer via can_receive(). So check for possible flushes on RCR changes. This will drop all pending packets on soft reset or disable which is the functional intent of the can_receive() logic.

---

 hw/net/smc91c111.c | 1 +
 1 file changed, 1 insertion(+)
hw/net/smc91c111.c +1 -0
index 8fc3deb..c19cdd1 100644
@@ -331,6 +331,7 @@ static void smc91c111_writeb(void *opaque, hwaddr offset,
             if (s->rcr & RCR_SOFT_RST) {
                 smc91c111_reset(DEVICE(s));
             }
+            smc91c111_flush_queued_packets(s);
             return;
         case 10: case 11: /* RPCR */
             /* Ignored */
Signature · 1 line
1.9.1

This should be useful for other NICs too.

Fam

#8 lore raw
In reply to Richard Purdie
Quoted 6 lines · Richard Purdie

Hi Peter,

Quoted 3 lines · Peter Crosthwaite

This should hopefully fix your bug, while addressing the extra concern I raised.

Quoted 8 lines
Quoted 4 lines

There was also inconsistent behaviour with corking packets through a soft reset which I notice and fixed.

Please let me know if this works for you.

I've run it through a few builds/tests in place of my own patches and so far it seems to be working, thanks!

Can we take that as a formal Tested-by? :)

Regards,
Peter

Quoted 4 lines

Cheers,

Richard

Tested-by
#9 lore raw
Quoted 11 lines · Peter Crosthwaite
Quoted 9 lines · Richard Purdie

Hi Peter,

Quoted 6 lines · Peter Crosthwaite

This should hopefully fix your bug, while addressing the extra concern I raised.

There was also inconsistent behaviour with corking packets through a soft reset which I notice and fixed.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Please let me know if this works for you.

I've run it through a few builds/tests in place of my own patches and so far it seems to be working, thanks!

Can we take that as a formal Tested-by? :)

if that helps :)

Cheers,

Richard

#10 lore raw
In reply to Peter Crosthwaite
Quoted 14 lines · Peter Crosthwaite

Hi Richard,

This should hopefully fix your bug, while addressing the extra concern I raised.

There was also inconsistent behaviour with corking packets through a soft reset which I notice and fixed.

Please let me know if this works for you.

Regards,
Peter

Quoted 12 lines

Peter Crosthwaite (3):
net: smc91c111: guard flush_queued_packets() on can_rx()
net: smc91c111: gate can_receive() on rx FIFO having a slot
net: smc91c111: flush packets on RCR register changes

 hw/net/smc91c111.c | 33 +++++++++++++++++++++++++--------
 1 file changed, 25 insertions(+), 8 deletions(-)
Signature · 1 line
1.9.1

Thanks, applied to my net tree:
https://github.com/stefanha/qemu/commits/net

Stefan