Skip to contentSkip to conversation

Fix MRU mismatch issue which may lead to data connection lost

PATCH In review Slark Xiao 2 participants 3 messages Jan 14, 2022 – Jan 15, 2022

Subsystems (3)
lore
Share & tools
lore

Slark Xiao

author
#0 lore raw

PATCH 1/1 Fix MRU mismatch issue which may lead to data connection lost

In pci_generic.c there is a 'mru_default' in struct mhi_pci_dev_info.
This value shall be used for whole mhi if it's given a value for a specific product. But in function mhi_net_rx_refill_work(), it's still using hard code value MHI_DEFAULT_MRU. 'mru_default' shall have higher priority than MHI_DEFAULT_MRU.
And after checking, this change could help fix a data connection lost issue.

---
 drivers/net/wwan/mhi_wwan_mbim.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/wwan/mhi_wwan_mbim.c +2 -2
index 71bf9b4f769f..6872782e8dd8 100644
@@ -385,13 +385,13 @@ static void mhi_net_rx_refill_work(struct work_struct *work)
 	int err;
 
 	while (!mhi_queue_is_full(mdev, DMA_FROM_DEVICE)) {
-		struct sk_buff *skb = alloc_skb(MHI_DEFAULT_MRU, GFP_KERNEL);
+		struct sk_buff *skb = alloc_skb(mbim->mru, GFP_KERNEL);
 
 		if (unlikely(!skb))
 			break;
 
 		err = mhi_queue_skb(mdev, DMA_FROM_DEVICE, skb,
-				    MHI_DEFAULT_MRU, MHI_EOT);
+				    mbim->mru, MHI_EOT);
 		if (unlikely(err)) {
 			kfree_skb(skb);
 			break;
Signature · 1 line
2.25.1
Reviewed-by on v1 1/1
#1 lore raw
Quoted 6 lines · Slark Xiao

In pci_generic.c there is a 'mru_default' in struct mhi_pci_dev_info.
This value shall be used for whole mhi if it's given a value for a specific product. But in function mhi_net_rx_refill_work(), it's still using hard code value MHI_DEFAULT_MRU. 'mru_default' shall have higher priority than MHI_DEFAULT_MRU.
And after checking, this change could help fix a data connection lost issue.

Interesting, not sure why it fixes data issues, since the device should comply with any size.Can you add a Fixes tag then? and add the correct [PATCH net] suffix in the subject:
https://www.kernel.org/doc/html/latest/networking/netdev-FAQ.html

With that:

Thanks,
Loic

Quoted 30 lines
---
 drivers/net/wwan/mhi_wwan_mbim.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/wwan/mhi_wwan_mbim.c +2 -3
index 71bf9b4f769f..6872782e8dd8 100644
@@ -385,13 +385,13 @@ static void mhi_net_rx_refill_work(struct work_struct *work)
        int err;
 
        while (!mhi_queue_is_full(mdev, DMA_FROM_DEVICE)) {
-               struct sk_buff *skb = alloc_skb(MHI_DEFAULT_MRU, GFP_KERNEL);
+               struct sk_buff *skb = alloc_skb(mbim->mru, GFP_KERNEL);
 
                if (unlikely(!skb))
                        break;
 
                err = mhi_queue_skb(mdev, DMA_FROM_DEVICE, skb,
-                                   MHI_DEFAULT_MRU, MHI_EOT);
+                                   mbim->mru, MHI_EOT);
                if (unlikely(err)) {
                        kfree_skb(skb);
                        break;
--

2.25.1

author
#2 lore raw
Quoted 13 lines · Loic Poulain
Quoted 6 lines · Slark Xiao

In pci_generic.c there is a 'mru_default' in struct mhi_pci_dev_info.
This value shall be used for whole mhi if it's given a value for a specific product. But in function mhi_net_rx_refill_work(), it's still using hard code value MHI_DEFAULT_MRU. 'mru_default' shall have higher priority than MHI_DEFAULT_MRU.
And after checking, this change could help fix a data connection lost issue.

Interesting, not sure why it fixes data issues, since the device should comply with any size.Can you add a Fixes tag then? and add the correct [PATCH net] suffix in the subject:
https://www.kernel.org/doc/html/latest/networking/netdev-FAQ.html

Quoted 7 lines

With that:

Thanks,
Loic

Hi Loic,
Actually, for our Foxconn device SDX55, it has been confirmed that MRU 3500
would lead to a IPA stuck. IPA stuck would make the data connection lost. For other MRU value, like 4000, 4096,32768, we can't reproduce it.
We also have report this issue in https://bugzilla.kernel.org/show_bug.cgi?id=215433 . I will send a new patch later.

Thanks
Slark

Quoted 33 lines
Quoted 33 lines
---
 drivers/net/wwan/mhi_wwan_mbim.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/wwan/mhi_wwan_mbim.c +2 -3
index 71bf9b4f769f..6872782e8dd8 100644
@@ -385,13 +385,13 @@ static void mhi_net_rx_refill_work(struct work_struct *work)
        int err;
 
        while (!mhi_queue_is_full(mdev, DMA_FROM_DEVICE)) {
-               struct sk_buff *skb = alloc_skb(MHI_DEFAULT_MRU, GFP_KERNEL);
+               struct sk_buff *skb = alloc_skb(mbim->mru, GFP_KERNEL);
 
                if (unlikely(!skb))
                        break;
 
                err = mhi_queue_skb(mdev, DMA_FROM_DEVICE, skb,
-                                   MHI_DEFAULT_MRU, MHI_EOT);
+                                   mbim->mru, MHI_EOT);
                if (unlikely(err)) {
                        kfree_skb(skb);
                        break;
--

2.25.1