* [PATCH 0/5] Drivers: hv: vmbus
@ 2016-03-19 1:13 K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
2016-03-23 19:43 ` [PATCH 0/5] Drivers: hv: vmbus KY Srinivasan
0 siblings, 2 replies; 14+ messages in thread
From: K. Y. Srinivasan @ 2016-03-19 1:13 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, olaf, apw, vkuznets, jasowang
Cc: K. Y. Srinivasan
Cleanup the Hyper-V ring buffer code. Also Implement APIs for
supporting copy-free operations on the read side.
K. Y. Srinivasan (5):
Drivers: hv: vmbus: Introduce functions for estimating room in the
ring buffer
Drivers: hv: vmbus: Use READ_ONCE() to read variables that are
volatile
Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
Drivers: hv: vmbus: Use the new virt_xx barrier code
Drivers: hv: vmbus: Implement copy-free read APIs
drivers/hv/ring_buffer.c | 99 ++++++++++++++++++++++++++++++++-------------
include/linux/hyperv.h | 33 +++++++++++++++
2 files changed, 103 insertions(+), 29 deletions(-)
--
1.7.4.1
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer
2016-03-19 1:13 [PATCH 0/5] Drivers: hv: vmbus K. Y. Srinivasan
@ 2016-03-19 1:14 ` K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 2/5] Drivers: hv: vmbus: Use READ_ONCE() to read variables that are volatile K. Y. Srinivasan
` (3 more replies)
2016-03-23 19:43 ` [PATCH 0/5] Drivers: hv: vmbus KY Srinivasan
1 sibling, 4 replies; 14+ messages in thread
From: K. Y. Srinivasan @ 2016-03-19 1:14 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, olaf, apw, vkuznets, jasowang
Cc: K. Y. Srinivasan
Introduce separate functions for estimating how much can be read from
and written to the ring buffer.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
drivers/hv/ring_buffer.c | 24 ++++--------------------
include/linux/hyperv.h | 27 +++++++++++++++++++++++++++
2 files changed, 31 insertions(+), 20 deletions(-)
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 085003a..902375b 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -38,8 +38,6 @@ void hv_begin_read(struct hv_ring_buffer_info *rbi)
u32 hv_end_read(struct hv_ring_buffer_info *rbi)
{
- u32 read;
- u32 write;
rbi->ring_buffer->interrupt_mask = 0;
mb();
@@ -49,9 +47,7 @@ u32 hv_end_read(struct hv_ring_buffer_info *rbi)
* If it is not, we raced and we need to process new
* incoming messages.
*/
- hv_get_ringbuffer_availbytes(rbi, &read, &write);
-
- return read;
+ return hv_get_bytes_to_read(rbi);
}
/*
@@ -106,18 +102,13 @@ static bool hv_need_to_signal(u32 old_write, struct hv_ring_buffer_info *rbi)
static bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
{
u32 cur_write_sz;
- u32 r_size;
- u32 write_loc = rbi->ring_buffer->write_index;
- u32 read_loc = rbi->ring_buffer->read_index;
u32 pending_sz = rbi->ring_buffer->pending_send_sz;
/* If the other end is not blocked on write don't bother. */
if (pending_sz == 0)
return false;
- r_size = rbi->ring_datasize;
- cur_write_sz = write_loc >= read_loc ? r_size - (write_loc - read_loc) :
- read_loc - write_loc;
+ cur_write_sz = hv_get_bytes_to_write(rbi);
if (cur_write_sz >= pending_sz)
return true;
@@ -317,7 +308,6 @@ int hv_ringbuffer_write(struct hv_ring_buffer_info *outring_info,
{
int i = 0;
u32 bytes_avail_towrite;
- u32 bytes_avail_toread;
u32 totalbytes_towrite = 0;
u32 next_write_location;
@@ -333,9 +323,7 @@ int hv_ringbuffer_write(struct hv_ring_buffer_info *outring_info,
if (lock)
spin_lock_irqsave(&outring_info->ring_lock, flags);
- hv_get_ringbuffer_availbytes(outring_info,
- &bytes_avail_toread,
- &bytes_avail_towrite);
+ bytes_avail_towrite = hv_get_bytes_to_write(outring_info);
/*
* If there is only room for the packet, assume it is full.
@@ -386,7 +374,6 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info *inring_info,
void *buffer, u32 buflen, u32 *buffer_actual_len,
u64 *requestid, bool *signal, bool raw)
{
- u32 bytes_avail_towrite;
u32 bytes_avail_toread;
u32 next_read_location = 0;
u64 prev_indices = 0;
@@ -402,10 +389,7 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info *inring_info,
*buffer_actual_len = 0;
*requestid = 0;
- hv_get_ringbuffer_availbytes(inring_info,
- &bytes_avail_toread,
- &bytes_avail_towrite);
-
+ bytes_avail_toread = hv_get_bytes_to_read(inring_info);
/* Make sure there is something to read */
if (bytes_avail_toread < sizeof(desc)) {
/*
diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
index ecd81c3..a6b053c 100644
--- a/include/linux/hyperv.h
+++ b/include/linux/hyperv.h
@@ -151,6 +151,33 @@ hv_get_ringbuffer_availbytes(struct hv_ring_buffer_info *rbi,
*read = dsize - *write;
}
+static inline u32 hv_get_bytes_to_read(struct hv_ring_buffer_info *rbi)
+{
+ u32 read_loc, write_loc, dsize, read;
+
+ dsize = rbi->ring_datasize;
+ read_loc = rbi->ring_buffer->read_index;
+ write_loc = READ_ONCE(rbi->ring_buffer->write_index);
+
+ read = write_loc >= read_loc ? (write_loc - read_loc) :
+ (dsize - read_loc) + write_loc;
+
+ return read;
+}
+
+static inline u32 hv_get_bytes_to_write(struct hv_ring_buffer_info *rbi)
+{
+ u32 read_loc, write_loc, dsize, write;
+
+ dsize = rbi->ring_datasize;
+ read_loc = READ_ONCE(rbi->ring_buffer->read_index);
+ write_loc = rbi->ring_buffer->write_index;
+
+ write = write_loc >= read_loc ? dsize - (write_loc - read_loc) :
+ read_loc - write_loc;
+ return write;
+}
+
/*
* VMBUS version is 32 bit entity broken up into
* two 16 bit quantities: major_number. minor_number.
--
1.7.4.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH 2/5] Drivers: hv: vmbus: Use READ_ONCE() to read variables that are volatile
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
@ 2016-03-19 1:14 ` K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read() K. Y. Srinivasan
` (2 subsequent siblings)
3 siblings, 0 replies; 14+ messages in thread
From: K. Y. Srinivasan @ 2016-03-19 1:14 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, olaf, apw, vkuznets, jasowang
Cc: K. Y. Srinivasan
Use the READ_ONCE macro to access variabes that can change asynchronously.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
drivers/hv/ring_buffer.c | 7 ++++---
1 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 902375b..2919395 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -69,7 +69,7 @@ u32 hv_end_read(struct hv_ring_buffer_info *rbi)
static bool hv_need_to_signal(u32 old_write, struct hv_ring_buffer_info *rbi)
{
mb();
- if (rbi->ring_buffer->interrupt_mask)
+ if (READ_ONCE(rbi->ring_buffer->interrupt_mask))
return false;
/* check interrupt_mask before read_index */
@@ -78,7 +78,7 @@ static bool hv_need_to_signal(u32 old_write, struct hv_ring_buffer_info *rbi)
* This is the only case we need to signal when the
* ring transitions from being empty to non-empty.
*/
- if (old_write == rbi->ring_buffer->read_index)
+ if (old_write == READ_ONCE(rbi->ring_buffer->read_index))
return true;
return false;
@@ -102,8 +102,9 @@ static bool hv_need_to_signal(u32 old_write, struct hv_ring_buffer_info *rbi)
static bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
{
u32 cur_write_sz;
- u32 pending_sz = rbi->ring_buffer->pending_send_sz;
+ u32 pending_sz;
+ pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
/* If the other end is not blocked on write don't bother. */
if (pending_sz == 0)
return false;
--
1.7.4.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 2/5] Drivers: hv: vmbus: Use READ_ONCE() to read variables that are volatile K. Y. Srinivasan
@ 2016-03-19 1:14 ` K. Y. Srinivasan
2016-03-21 8:19 ` Vitaly Kuznetsov
2016-03-19 1:14 ` [PATCH 4/5] Drivers: hv: vmbus: Use the new virt_xx barrier code K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs K. Y. Srinivasan
3 siblings, 1 reply; 14+ messages in thread
From: K. Y. Srinivasan @ 2016-03-19 1:14 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, olaf, apw, vkuznets, jasowang
Cc: K. Y. Srinivasan, stable
We need to issue a full memory barrier prior making a signalling decision.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Cc: stable@vger.kernel.org
---
drivers/hv/ring_buffer.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 2919395..67dc245 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -104,6 +104,7 @@ static bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
u32 cur_write_sz;
u32 pending_sz;
+ mb();
pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
/* If the other end is not blocked on write don't bother. */
if (pending_sz == 0)
--
1.7.4.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH 4/5] Drivers: hv: vmbus: Use the new virt_xx barrier code
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 2/5] Drivers: hv: vmbus: Use READ_ONCE() to read variables that are volatile K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read() K. Y. Srinivasan
@ 2016-03-19 1:14 ` K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs K. Y. Srinivasan
3 siblings, 0 replies; 14+ messages in thread
From: K. Y. Srinivasan @ 2016-03-19 1:14 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, olaf, apw, vkuznets, jasowang
Cc: K. Y. Srinivasan
Use the virt_xx barriers that have been defined for use in virtual machines.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
drivers/hv/ring_buffer.c | 14 +++++++-------
1 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index 67dc245..c2c2b2e 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -33,14 +33,14 @@
void hv_begin_read(struct hv_ring_buffer_info *rbi)
{
rbi->ring_buffer->interrupt_mask = 1;
- mb();
+ virt_mb();
}
u32 hv_end_read(struct hv_ring_buffer_info *rbi)
{
rbi->ring_buffer->interrupt_mask = 0;
- mb();
+ virt_mb();
/*
* Now check to see if the ring buffer is still empty.
@@ -68,12 +68,12 @@ u32 hv_end_read(struct hv_ring_buffer_info *rbi)
static bool hv_need_to_signal(u32 old_write, struct hv_ring_buffer_info *rbi)
{
- mb();
+ virt_mb();
if (READ_ONCE(rbi->ring_buffer->interrupt_mask))
return false;
/* check interrupt_mask before read_index */
- rmb();
+ virt_rmb();
/*
* This is the only case we need to signal when the
* ring transitions from being empty to non-empty.
@@ -104,7 +104,7 @@ static bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
u32 cur_write_sz;
u32 pending_sz;
- mb();
+ virt_mb();
pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
/* If the other end is not blocked on write don't bother. */
if (pending_sz == 0)
@@ -359,7 +359,7 @@ int hv_ringbuffer_write(struct hv_ring_buffer_info *outring_info,
sizeof(u64));
/* Issue a full memory barrier before updating the write index */
- mb();
+ virt_mb();
/* Now, update the write location */
hv_set_next_write_location(outring_info, next_write_location);
@@ -435,7 +435,7 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info *inring_info,
* the writer may start writing to the read area once the read index
* is updated.
*/
- mb();
+ virt_mb();
/* Update the read index */
hv_set_next_read_location(inring_info, next_read_location);
--
1.7.4.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
` (2 preceding siblings ...)
2016-03-19 1:14 ` [PATCH 4/5] Drivers: hv: vmbus: Use the new virt_xx barrier code K. Y. Srinivasan
@ 2016-03-19 1:14 ` K. Y. Srinivasan
2016-03-21 8:29 ` Vitaly Kuznetsov
3 siblings, 1 reply; 14+ messages in thread
From: K. Y. Srinivasan @ 2016-03-19 1:14 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, olaf, apw, vkuznets, jasowang
Cc: K. Y. Srinivasan
Implement copy-free read APIs.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
---
drivers/hv/ring_buffer.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++
include/linux/hyperv.h | 6 +++++
2 files changed, 61 insertions(+), 0 deletions(-)
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index c2c2b2e..c80e1f3 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -444,3 +444,58 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info *inring_info,
return ret;
}
+
+/*
+ * In-place read functions.
+ */
+bool get_next_pkt_raw(struct vmbus_channel *channel,
+ struct vmpacket_descriptor **desc)
+{
+ struct hv_ring_buffer_info *ring_info = &channel->inbound;
+ u32 read_loc = ring_info->ring_buffer->read_index;
+ void *ring_buffer = hv_get_ring_buffer(ring_info);
+ struct vmpacket_descriptor *cur_desc;
+ u32 packetlen;
+ u32 dsize = ring_info->ring_datasize;
+ u32 bytes_avail_toread = hv_get_bytes_to_read(ring_info);
+
+ if (bytes_avail_toread < sizeof(struct vmpacket_descriptor))
+ return false;
+
+ if ((read_loc + sizeof(*desc)) > dsize)
+ return false;
+
+ cur_desc = ring_buffer + read_loc;
+ packetlen = cur_desc->len8 << 3;
+
+ if ((read_loc + packetlen + 8) > (dsize - 1))
+ return false;
+
+ *desc = cur_desc;
+ return true;
+}
+EXPORT_SYMBOL_GPL(get_next_pkt_raw);
+
+void put_pkt_raw(struct vmbus_channel *channel,
+ struct vmpacket_descriptor *desc)
+{
+ struct hv_ring_buffer_info *ring_info = &channel->inbound;
+ u32 read_loc = ring_info->ring_buffer->read_index;
+ u32 packetlen = desc->len8 << 3;
+ u32 dsize = ring_info->ring_datasize;
+
+ if ((read_loc + packetlen + 8) > dsize)
+ BUG();
+
+ /*
+ * Make sure all reads are done before we update the read index since
+ * the writer may start writing to the read area once the read index
+ * is updated.
+ */
+ virt_mb();
+ ring_info->ring_buffer->read_index += packetlen + 8;
+
+ if (hv_need_to_signal_on_read(ring_info))
+ vmbus_set_event(channel);
+}
+EXPORT_SYMBOL_GPL(put_pkt_raw);
diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
index a6b053c..455f3f0 100644
--- a/include/linux/hyperv.h
+++ b/include/linux/hyperv.h
@@ -1035,6 +1035,12 @@ extern int vmbus_recvpacket_raw(struct vmbus_channel *channel,
u32 *buffer_actual_len,
u64 *requestid);
+bool get_next_pkt_raw(struct vmbus_channel *channel,
+ struct vmpacket_descriptor **desc);
+
+void put_pkt_raw(struct vmbus_channel *channel,
+ struct vmpacket_descriptor *desc);
+
extern void vmbus_ontimer(unsigned long data);
--
1.7.4.1
^ permalink raw reply related [flat|nested] 14+ messages in thread
* Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
2016-03-19 1:14 ` [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read() K. Y. Srinivasan
@ 2016-03-21 8:19 ` Vitaly Kuznetsov
2016-03-22 0:04 ` KY Srinivasan
0 siblings, 1 reply; 14+ messages in thread
From: Vitaly Kuznetsov @ 2016-03-21 8:19 UTC (permalink / raw)
To: K. Y. Srinivasan; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang, stable
"K. Y. Srinivasan" <kys@microsoft.com> writes:
> We need to issue a full memory barrier prior making a signalling
> decision.
Any reason this should be mb()? This is pretty strong and will probably
lead to performace regression ... and, btw, we have another mb() in
hv_ringbuffer_read().
Could you please describe the scenarion you're trying to protect against
so we could search for a better solution?
>
> Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
> Cc: stable@vger.kernel.org
> ---
> drivers/hv/ring_buffer.c | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
> index 2919395..67dc245 100644
> --- a/drivers/hv/ring_buffer.c
> +++ b/drivers/hv/ring_buffer.c
> @@ -104,6 +104,7 @@ static bool hv_need_to_signal_on_read(struct hv_ring_buffer_info *rbi)
> u32 cur_write_sz;
> u32 pending_sz;
>
> + mb();
> pending_sz = READ_ONCE(rbi->ring_buffer->pending_send_sz);
> /* If the other end is not blocked on write don't bother. */
> if (pending_sz == 0)
--
Vitaly
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs
2016-03-19 1:14 ` [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs K. Y. Srinivasan
@ 2016-03-21 8:29 ` Vitaly Kuznetsov
2016-03-21 23:42 ` KY Srinivasan
0 siblings, 1 reply; 14+ messages in thread
From: Vitaly Kuznetsov @ 2016-03-21 8:29 UTC (permalink / raw)
To: K. Y. Srinivasan; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang
"K. Y. Srinivasan" <kys@microsoft.com> writes:
> Implement copy-free read APIs.
>
> Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
> ---
> drivers/hv/ring_buffer.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++
> include/linux/hyperv.h | 6 +++++
> 2 files changed, 61 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
> index c2c2b2e..c80e1f3 100644
> --- a/drivers/hv/ring_buffer.c
> +++ b/drivers/hv/ring_buffer.c
> @@ -444,3 +444,58 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info *inring_info,
>
> return ret;
> }
> +
> +/*
> + * In-place read functions.
> + */
> +bool get_next_pkt_raw(struct vmbus_channel *channel,
> + struct vmpacket_descriptor **desc)
I suggest we make these functions return pointers and use
ERR_PTR()/IS_ERR() when we need to return an error.
> +{
> + struct hv_ring_buffer_info *ring_info = &channel->inbound;
> + u32 read_loc = ring_info->ring_buffer->read_index;
> + void *ring_buffer = hv_get_ring_buffer(ring_info);
> + struct vmpacket_descriptor *cur_desc;
> + u32 packetlen;
> + u32 dsize = ring_info->ring_datasize;
> + u32 bytes_avail_toread = hv_get_bytes_to_read(ring_info);
> +
> + if (bytes_avail_toread < sizeof(struct vmpacket_descriptor))
> + return false;
> +
> + if ((read_loc + sizeof(*desc)) > dsize)
> + return false;
> +
> + cur_desc = ring_buffer + read_loc;
> + packetlen = cur_desc->len8 << 3;
> +
> + if ((read_loc + packetlen + 8) > (dsize - 1))
Could you please add a define for this '8'?
> + return false;
> +
> + *desc = cur_desc;
> + return true;
> +}
> +EXPORT_SYMBOL_GPL(get_next_pkt_raw);
> +
> +void put_pkt_raw(struct vmbus_channel *channel,
> + struct vmpacket_descriptor *desc)
> +{
> + struct hv_ring_buffer_info *ring_info = &channel->inbound;
> + u32 read_loc = ring_info->ring_buffer->read_index;
> + u32 packetlen = desc->len8 << 3;
> + u32 dsize = ring_info->ring_datasize;
> +
> + if ((read_loc + packetlen + 8) > dsize)
> + BUG();
> +
> + /*
> + * Make sure all reads are done before we update the read index since
> + * the writer may start writing to the read area once the read index
> + * is updated.
> + */
> + virt_mb();
> + ring_info->ring_buffer->read_index += packetlen + 8;
> +
> + if (hv_need_to_signal_on_read(ring_info))
> + vmbus_set_event(channel);
> +}
> +EXPORT_SYMBOL_GPL(put_pkt_raw);
> diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
> index a6b053c..455f3f0 100644
> --- a/include/linux/hyperv.h
> +++ b/include/linux/hyperv.h
> @@ -1035,6 +1035,12 @@ extern int vmbus_recvpacket_raw(struct vmbus_channel *channel,
> u32 *buffer_actual_len,
> u64 *requestid);
>
> +bool get_next_pkt_raw(struct vmbus_channel *channel,
> + struct vmpacket_descriptor **desc);
> +
> +void put_pkt_raw(struct vmbus_channel *channel,
> + struct vmpacket_descriptor *desc);
> +
>
> extern void vmbus_ontimer(unsigned long data);
--
Vitaly
^ permalink raw reply [flat|nested] 14+ messages in thread
* RE: [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs
2016-03-21 8:29 ` Vitaly Kuznetsov
@ 2016-03-21 23:42 ` KY Srinivasan
0 siblings, 0 replies; 14+ messages in thread
From: KY Srinivasan @ 2016-03-21 23:42 UTC (permalink / raw)
To: Vitaly Kuznetsov; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang
> -----Original Message-----
> From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
> Sent: Monday, March 21, 2016 1:29 AM
> To: KY Srinivasan <kys@microsoft.com>
> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> jasowang@redhat.com
> Subject: Re: [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs
>
> "K. Y. Srinivasan" <kys@microsoft.com> writes:
>
> > Implement copy-free read APIs.
> >
> > Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
> > ---
> > drivers/hv/ring_buffer.c | 55
> ++++++++++++++++++++++++++++++++++++++++++++++
> > include/linux/hyperv.h | 6 +++++
> > 2 files changed, 61 insertions(+), 0 deletions(-)
> >
> > diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
> > index c2c2b2e..c80e1f3 100644
> > --- a/drivers/hv/ring_buffer.c
> > +++ b/drivers/hv/ring_buffer.c
> > @@ -444,3 +444,58 @@ int hv_ringbuffer_read(struct hv_ring_buffer_info
> *inring_info,
> >
> > return ret;
> > }
> > +
> > +/*
> > + * In-place read functions.
> > + */
> > +bool get_next_pkt_raw(struct vmbus_channel *channel,
> > + struct vmpacket_descriptor **desc)
>
> I suggest we make these functions return pointers and use
> ERR_PTR()/IS_ERR() when we need to return an error.
Either we get a pointer to a descriptor or not; there is no other
error here. I will change the signature accordingly.
>
> > +{
> > + struct hv_ring_buffer_info *ring_info = &channel->inbound;
> > + u32 read_loc = ring_info->ring_buffer->read_index;
> > + void *ring_buffer = hv_get_ring_buffer(ring_info);
> > + struct vmpacket_descriptor *cur_desc;
> > + u32 packetlen;
> > + u32 dsize = ring_info->ring_datasize;
> > + u32 bytes_avail_toread = hv_get_bytes_to_read(ring_info);
> > +
> > + if (bytes_avail_toread < sizeof(struct vmpacket_descriptor))
> > + return false;
> > +
> > + if ((read_loc + sizeof(*desc)) > dsize)
> > + return false;
> > +
> > + cur_desc = ring_buffer + read_loc;
> > + packetlen = cur_desc->len8 << 3;
> > +
> > + if ((read_loc + packetlen + 8) > (dsize - 1))
>
> Could you please add a define for this '8'?
Sure.
K. Y
^ permalink raw reply [flat|nested] 14+ messages in thread
* RE: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
2016-03-21 8:19 ` Vitaly Kuznetsov
@ 2016-03-22 0:04 ` KY Srinivasan
2016-03-22 9:55 ` Vitaly Kuznetsov
0 siblings, 1 reply; 14+ messages in thread
From: KY Srinivasan @ 2016-03-22 0:04 UTC (permalink / raw)
To: Vitaly Kuznetsov; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang, stable
> -----Original Message-----
> From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
> Sent: Monday, March 21, 2016 1:19 AM
> To: KY Srinivasan <kys@microsoft.com>
> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> jasowang@redhat.com; stable@vger.kernel.org
> Subject: Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
> hv_need_to_signal_on_read()
>
> "K. Y. Srinivasan" <kys@microsoft.com> writes:
>
> > We need to issue a full memory barrier prior making a signalling
> > decision.
>
> Any reason this should be mb()? This is pretty strong and will probably
> lead to performace regression ... and, btw, we have another mb() in
> hv_ringbuffer_read().
>
> Could you please describe the scenarion you're trying to protect against
> so we could search for a better solution?
If the reading of the pend_sz (in the function hv_need_to_signal_on_read)
were to be reordered and read before we commit the new read index we could
have a problem. If the host were to set the pending_sz after we have sampled pending_sz
and go to sleep before we commit the read index, we could miss sending the interrupt.
Regards,
K. Y
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
2016-03-22 0:04 ` KY Srinivasan
@ 2016-03-22 9:55 ` Vitaly Kuznetsov
2016-03-22 14:37 ` KY Srinivasan
2016-03-22 17:11 ` KY Srinivasan
0 siblings, 2 replies; 14+ messages in thread
From: Vitaly Kuznetsov @ 2016-03-22 9:55 UTC (permalink / raw)
To: KY Srinivasan; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang, stable
KY Srinivasan <kys@microsoft.com> writes:
>> -----Original Message-----
>> From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
>> Sent: Monday, March 21, 2016 1:19 AM
>> To: KY Srinivasan <kys@microsoft.com>
>> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
>> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
>> jasowang@redhat.com; stable@vger.kernel.org
>> Subject: Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
>> hv_need_to_signal_on_read()
>>
>> "K. Y. Srinivasan" <kys@microsoft.com> writes:
>>
>> > We need to issue a full memory barrier prior making a signalling
>> > decision.
>>
>> Any reason this should be mb()? This is pretty strong and will probably
>> lead to performace regression ... and, btw, we have another mb() in
>> hv_ringbuffer_read().
>>
>> Could you please describe the scenarion you're trying to protect against
>> so we could search for a better solution?
>
> If the reading of the pend_sz (in the function hv_need_to_signal_on_read)
> were to be reordered and read before we commit the new read index we could
> have a problem.
If these are two reads we can add a lightweight barrier just preventing
compiler from reordering (e.g. smp_rmb()), right?
> If the host were to set the pending_sz after we have sampled pending_sz
> and go to sleep before we commit the read index, we could miss sending
> the interrupt.
so write and then we read and we need to prevent reordering... not sure
how to get rid on mb() then ...
--
Vitaly
^ permalink raw reply [flat|nested] 14+ messages in thread
* RE: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
2016-03-22 9:55 ` Vitaly Kuznetsov
@ 2016-03-22 14:37 ` KY Srinivasan
2016-03-22 17:11 ` KY Srinivasan
1 sibling, 0 replies; 14+ messages in thread
From: KY Srinivasan @ 2016-03-22 14:37 UTC (permalink / raw)
To: Vitaly Kuznetsov; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang, stable
> -----Original Message-----
> From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
> Sent: Tuesday, March 22, 2016 2:56 AM
> To: KY Srinivasan <kys@microsoft.com>
> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> jasowang@redhat.com; stable@vger.kernel.org
> Subject: Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
> hv_need_to_signal_on_read()
>
> KY Srinivasan <kys@microsoft.com> writes:
>
> >> -----Original Message-----
> >> From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
> >> Sent: Monday, March 21, 2016 1:19 AM
> >> To: KY Srinivasan <kys@microsoft.com>
> >> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> >> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> >> jasowang@redhat.com; stable@vger.kernel.org
> >> Subject: Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
> >> hv_need_to_signal_on_read()
> >>
> >> "K. Y. Srinivasan" <kys@microsoft.com> writes:
> >>
> >> > We need to issue a full memory barrier prior making a signalling
> >> > decision.
> >>
> >> Any reason this should be mb()? This is pretty strong and will probably
> >> lead to performace regression ... and, btw, we have another mb() in
> >> hv_ringbuffer_read().
> >>
> >> Could you please describe the scenarion you're trying to protect against
> >> so we could search for a better solution?
> >
> > If the reading of the pend_sz (in the function
> hv_need_to_signal_on_read)
> > were to be reordered and read before we commit the new read index we
> could
> > have a problem.
>
>
> If these are two reads we can add a lightweight barrier just preventing
> compiler from reordering (e.g. smp_rmb()), right?
>
> > If the host were to set the pending_sz after we have sampled pending_sz
> > and go to sleep before we commit the read index, we could miss sending
> > the interrupt.
>
> so write and then we read and we need to prevent reordering... not sure
> how to get rid on mb() then ...
The other memory barrier in the function (prior to writing the read index)
has been there forever and I am not sure why that needs to be a full barrier.
I feel a read barrier should suffice.
Regards,
K. Y
>
> --
> Vitaly
^ permalink raw reply [flat|nested] 14+ messages in thread
* RE: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
2016-03-22 9:55 ` Vitaly Kuznetsov
2016-03-22 14:37 ` KY Srinivasan
@ 2016-03-22 17:11 ` KY Srinivasan
1 sibling, 0 replies; 14+ messages in thread
From: KY Srinivasan @ 2016-03-22 17:11 UTC (permalink / raw)
To: Vitaly Kuznetsov; +Cc: gregkh, linux-kernel, devel, olaf, apw, jasowang, stable
> -----Original Message-----
> From: KY Srinivasan
> Sent: Tuesday, March 22, 2016 7:37 AM
> To: 'Vitaly Kuznetsov' <vkuznets@redhat.com>
> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> jasowang@redhat.com; stable@vger.kernel.org
> Subject: RE: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
> hv_need_to_signal_on_read()
>
>
>
> > -----Original Message-----
> > From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
> > Sent: Tuesday, March 22, 2016 2:56 AM
> > To: KY Srinivasan <kys@microsoft.com>
> > Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> > devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> > jasowang@redhat.com; stable@vger.kernel.org
> > Subject: Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
> > hv_need_to_signal_on_read()
> >
> > KY Srinivasan <kys@microsoft.com> writes:
> >
> > >> -----Original Message-----
> > >> From: Vitaly Kuznetsov [mailto:vkuznets@redhat.com]
> > >> Sent: Monday, March 21, 2016 1:19 AM
> > >> To: KY Srinivasan <kys@microsoft.com>
> > >> Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> > >> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> > >> jasowang@redhat.com; stable@vger.kernel.org
> > >> Subject: Re: [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in
> > >> hv_need_to_signal_on_read()
> > >>
> > >> "K. Y. Srinivasan" <kys@microsoft.com> writes:
> > >>
> > >> > We need to issue a full memory barrier prior making a signalling
> > >> > decision.
> > >>
> > >> Any reason this should be mb()? This is pretty strong and will probably
> > >> lead to performace regression ... and, btw, we have another mb() in
> > >> hv_ringbuffer_read().
> > >>
> > >> Could you please describe the scenarion you're trying to protect against
> > >> so we could search for a better solution?
> > >
> > > If the reading of the pend_sz (in the function
> > hv_need_to_signal_on_read)
> > > were to be reordered and read before we commit the new read index
> we
> > could
> > > have a problem.
> >
> >
> > If these are two reads we can add a lightweight barrier just preventing
> > compiler from reordering (e.g. smp_rmb()), right?
> >
> > > If the host were to set the pending_sz after we have sampled
> pending_sz
> > > and go to sleep before we commit the read index, we could miss sending
> > > the interrupt.
> >
> > so write and then we read and we need to prevent reordering... not sure
> > how to get rid on mb() then ...
>
> The other memory barrier in the function (prior to writing the read index)
> has been there forever and I am not sure why that needs to be a full barrier.
> I feel a read barrier should suffice.
I may also look at restructuring these APIs to not always check for signaling.
I will experiment with that scheme to minimize the barrier calls.
K. Y
^ permalink raw reply [flat|nested] 14+ messages in thread
* RE: [PATCH 0/5] Drivers: hv: vmbus
2016-03-19 1:13 [PATCH 0/5] Drivers: hv: vmbus K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
@ 2016-03-23 19:43 ` KY Srinivasan
1 sibling, 0 replies; 14+ messages in thread
From: KY Srinivasan @ 2016-03-23 19:43 UTC (permalink / raw)
To: KY Srinivasan, gregkh, linux-kernel, devel, olaf, apw, vkuznets,
jasowang
> -----Original Message-----
> From: K. Y. Srinivasan [mailto:kys@microsoft.com]
> Sent: Friday, March 18, 2016 6:14 PM
> To: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> devel@linuxdriverproject.org; olaf@aepfle.de; apw@canonical.com;
> vkuznets@redhat.com; jasowang@redhat.com
> Cc: KY Srinivasan <kys@microsoft.com>
> Subject: [PATCH 0/5] Drivers: hv: vmbus
>
> Cleanup the Hyper-V ring buffer code. Also Implement APIs for
> supporting copy-free operations on the read side.
>
> K. Y. Srinivasan (5):
> Drivers: hv: vmbus: Introduce functions for estimating room in the
> ring buffer
> Drivers: hv: vmbus: Use READ_ONCE() to read variables that are
> volatile
> Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read()
> Drivers: hv: vmbus: Use the new virt_xx barrier code
> Drivers: hv: vmbus: Implement copy-free read APIs
>
> drivers/hv/ring_buffer.c | 99 ++++++++++++++++++++++++++++++++-----
> --------
> include/linux/hyperv.h | 33 +++++++++++++++
> 2 files changed, 103 insertions(+), 29 deletions(-)
Greg,
Please drop this series. I am going to send a new patch-set and looking at the changes I
have made, I feel it may not be appropriate to simply bump the version.
Regards,
K. Y
>
> --
> 1.7.4.1
^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2016-03-23 19:43 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-03-19 1:13 [PATCH 0/5] Drivers: hv: vmbus K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 1/5] Drivers: hv: vmbus: Introduce functions for estimating room in the ring buffer K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 2/5] Drivers: hv: vmbus: Use READ_ONCE() to read variables that are volatile K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 3/5] Drivers: hv: vmbus: Fix a bug in hv_need_to_signal_on_read() K. Y. Srinivasan
2016-03-21 8:19 ` Vitaly Kuznetsov
2016-03-22 0:04 ` KY Srinivasan
2016-03-22 9:55 ` Vitaly Kuznetsov
2016-03-22 14:37 ` KY Srinivasan
2016-03-22 17:11 ` KY Srinivasan
2016-03-19 1:14 ` [PATCH 4/5] Drivers: hv: vmbus: Use the new virt_xx barrier code K. Y. Srinivasan
2016-03-19 1:14 ` [PATCH 5/5] Drivers: hv: vmbus: Implement copy-free read APIs K. Y. Srinivasan
2016-03-21 8:29 ` Vitaly Kuznetsov
2016-03-21 23:42 ` KY Srinivasan
2016-03-23 19:43 ` [PATCH 0/5] Drivers: hv: vmbus KY Srinivasan
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.