Message ID | 151811766130.10712.18293368656209944798.email-sent-by-dnelson@aqua |
---|---|
State | Accepted, archived |
Delegated to: | David Miller |
Headers | show |
Series | net: thunder: change q_len's type to handle max ring size | expand |
From: Dean Nelson <dnelson@redhat.com> Date: > The Cavium thunder nicvf driver supports rx/tx rings of up to 65536 entries per. > The number of entires are stored in the q_len member of struct q_desc_mem. The > problem is that q_len being a u16, results in 65536 becoming 0. > > In getting pointers to descriptors in the rings, the driver uses q_len minus 1 > as a mask after incrementing the pointer, in order to go back to the beginning > and not go past the end of the ring. > > With the q_len set to 0 the mask is no longer correct and the driver does go > beyond the end of the ring, causing various ills. Usually the first thing that > shows up is a "NETDEV WATCHDOG: enP2p1s0f1 (nicvf): transmit queue 7 timed out" > warning. > > This patch remedies the problem by changing q_len to a u32. > > Signed-off-by: Dean Nelson <dnelson@redhat.com> Applied, thanks. Another way to solve this could have been to encode that length as "length - 1"
On 02/08/2018 02:34 PM, David Miller wrote: > From: Dean Nelson <dnelson@redhat.com> > Date: > >> The Cavium thunder nicvf driver supports rx/tx rings of up to 65536 entries per. >> The number of entires are stored in the q_len member of struct q_desc_mem. The >> problem is that q_len being a u16, results in 65536 becoming 0. >> >> In getting pointers to descriptors in the rings, the driver uses q_len minus 1 >> as a mask after incrementing the pointer, in order to go back to the beginning >> and not go past the end of the ring. >> >> With the q_len set to 0 the mask is no longer correct and the driver does go >> beyond the end of the ring, causing various ills. Usually the first thing that >> shows up is a "NETDEV WATCHDOG: enP2p1s0f1 (nicvf): transmit queue 7 timed out" >> warning. >> >> This patch remedies the problem by changing q_len to a u32. >> >> Signed-off-by: Dean Nelson <dnelson@redhat.com> > > Applied, thanks. Thank you! > > Another way to solve this could have been to encode that length > as "length - 1" True. I had pondered that, but felt that since changing q_len's type didn't add any length to the structure and that it was less impactful from a number-of-lines of code changed perspective, I'd opt for this route. Cavium, if you'd prefer this goes the route that Dave just mentioned, please let me know and I can make a new patch against what's been applied? Thanks, Dean
On Fri, Feb 9, 2018 at 3:27 AM, Dean Nelson <dnelson@redhat.com> wrote: > On 02/08/2018 02:34 PM, David Miller wrote: >> >> From: Dean Nelson <dnelson@redhat.com> >> Date: >> >>> The Cavium thunder nicvf driver supports rx/tx rings of up to 65536 >>> entries per. >>> The number of entires are stored in the q_len member of struct >>> q_desc_mem. The >>> problem is that q_len being a u16, results in 65536 becoming 0. >>> >>> In getting pointers to descriptors in the rings, the driver uses q_len >>> minus 1 >>> as a mask after incrementing the pointer, in order to go back to the >>> beginning >>> and not go past the end of the ring. >>> >>> With the q_len set to 0 the mask is no longer correct and the driver does >>> go >>> beyond the end of the ring, causing various ills. Usually the first thing >>> that >>> shows up is a "NETDEV WATCHDOG: enP2p1s0f1 (nicvf): transmit queue 7 >>> timed out" >>> warning. >>> >>> This patch remedies the problem by changing q_len to a u32. >>> >>> Signed-off-by: Dean Nelson <dnelson@redhat.com> >> >> >> Applied, thanks. > > > Thank you! > >> >> Another way to solve this could have been to encode that length >> as "length - 1" > > > True. I had pondered that, but felt that since changing q_len's type > didn't add any length to the structure and that it was less impactful > from a number-of-lines of code changed perspective, I'd opt for this > route. > > Cavium, if you'd prefer this goes the route that Dave just mentioned, > please let me know and I can make a new patch against what's been > applied? Thanks for fixing this and i think the current patch is fine. Thanks, Sunil. > > Thanks, > Dean > > > > > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On 02/08/2018 10:29 PM, Sunil Kovvuri wrote: > On Fri, Feb 9, 2018 at 3:27 AM, Dean Nelson <dnelson@redhat.com> wrote: >> On 02/08/2018 02:34 PM, David Miller wrote: >>> >>> From: Dean Nelson <dnelson@redhat.com> >>> Date: >>> >>>> The Cavium thunder nicvf driver supports rx/tx rings of up to 65536 >>>> entries per. ... >>> >>> Another way to solve this could have been to encode that length >>> as "length - 1" >> >> >> True. I had pondered that, but felt that since changing q_len's type >> didn't add any length to the structure and that it was less impactful >> from a number-of-lines of code changed perspective, I'd opt for this >> route. >> >> Cavium, if you'd prefer this goes the route that Dave just mentioned, >> please let me know and I can make a new patch against what's been >> applied? > > Thanks for fixing this and i think the current patch is fine. You're welcome. And thanks for responding. So I'll leave things as they are.
diff --git a/drivers/net/ethernet/cavium/thunder/nicvf_queues.h b/drivers/net/ethernet/cavium/thunder/nicvf_queues.h index 7d1e4e2aaad0..ce1eed7a6d63 100644 --- a/drivers/net/ethernet/cavium/thunder/nicvf_queues.h +++ b/drivers/net/ethernet/cavium/thunder/nicvf_queues.h @@ -213,7 +213,7 @@ struct rx_tx_queue_stats { struct q_desc_mem { dma_addr_t dma; u64 size; - u16 q_len; + u32 q_len; dma_addr_t phys_base; void *base; void *unalign_base;
The Cavium thunder nicvf driver supports rx/tx rings of up to 65536 entries per. The number of entires are stored in the q_len member of struct q_desc_mem. The problem is that q_len being a u16, results in 65536 becoming 0. In getting pointers to descriptors in the rings, the driver uses q_len minus 1 as a mask after incrementing the pointer, in order to go back to the beginning and not go past the end of the ring. With the q_len set to 0 the mask is no longer correct and the driver does go beyond the end of the ring, causing various ills. Usually the first thing that shows up is a "NETDEV WATCHDOG: enP2p1s0f1 (nicvf): transmit queue 7 timed out" warning. This patch remedies the problem by changing q_len to a u32. Signed-off-by: Dean Nelson <dnelson@redhat.com> --- drivers/net/ethernet/cavium/thunder/nicvf_queues.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)