Marvell Wifi PCIe modules don't always behave nicely for PCIe power
management when their firmware hasn't been loaded, particularly after
suspending the PCIe link one or more times. When this happens, we might
end up spinning forever in this status-polling tight loop. Let's make
this less tight by adding a timeout and by sleeping a bit in between
reads, as we do with the other similar loops.
This prevents us from hogging a CPU even in such pathological cases, and
allows the FW initialization to just fail gracefully instead.
I chose the same polling parameters as the earlier loop in this
function, and empirically, I found that this loop never makes it more
than about 12 cycles in a sane FW init sequence. I had no official
information on the actual intended latency for this portion of the
download.
Signed-off-by: Brian Norris <[email protected]>
---
drivers/net/wireless/marvell/mwifiex/pcie.c | 16 +++++++++++++---
1 file changed, 13 insertions(+), 3 deletions(-)
diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c b/drivers/net/wireless/marvell/mwifiex/pcie.c
index 4b89f557d0b6..9f9ea1350591 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie.c
+++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
@@ -2050,7 +2050,7 @@ static int mwifiex_prog_fw_w_helper(struct mwifiex_adapter *adapter,
}
/* Wait for the command done interrupt */
- do {
+ for (tries = 0; tries < MAX_POLL_TRIES; tries++) {
if (mwifiex_read_reg(adapter, PCIE_CPU_INT_STATUS,
&ireg_intr)) {
mwifiex_dbg(adapter, ERROR,
@@ -2062,8 +2062,18 @@ static int mwifiex_prog_fw_w_helper(struct mwifiex_adapter *adapter,
ret = -1;
goto done;
}
- } while ((ireg_intr & CPU_INTR_DOOR_BELL) ==
- CPU_INTR_DOOR_BELL);
+ if (!(ireg_intr & CPU_INTR_DOOR_BELL))
+ break;
+ usleep_range(10, 20);
+ }
+ if (ireg_intr & CPU_INTR_DOOR_BELL) {
+ mwifiex_dbg(adapter, ERROR, "%s: Card failed to ACK download\n",
+ __func__);
+ mwifiex_unmap_pci_memory(adapter, skb,
+ PCI_DMA_TODEVICE);
+ ret = -1;
+ goto done;
+ }
mwifiex_unmap_pci_memory(adapter, skb, PCI_DMA_TODEVICE);
--
2.8.0.rc3.226.g39d4020
> From: Brian Norris [mailto:[email protected]]
> Sent: Wednesday, November 23, 2016 8:10 AM
> To: Amitkumar Karwar; Nishant Sarmukadam
> Cc: [email protected]; Kalle Valo; linux-
> [email protected]; Cathy Luo; Dmitry Torokhov; Brian Norris
> Subject: [PATCH] mwifiex: pcie: implement timeout loop for FW
> programming doorbell
>
> Marvell Wifi PCIe modules don't always behave nicely for PCIe power
> management when their firmware hasn't been loaded, particularly after
> suspending the PCIe link one or more times. When this happens, we might
> end up spinning forever in this status-polling tight loop. Let's make
> this less tight by adding a timeout and by sleeping a bit in between
> reads, as we do with the other similar loops.
>
> This prevents us from hogging a CPU even in such pathological cases,
> and allows the FW initialization to just fail gracefully instead.
>
> I chose the same polling parameters as the earlier loop in this
> function, and empirically, I found that this loop never makes it more
> than about 12 cycles in a sane FW init sequence. I had no official
> information on the actual intended latency for this portion of the
> download.
>
> Signed-off-by: Brian Norris <[email protected]>
> ---
> drivers/net/wireless/marvell/mwifiex/pcie.c | 16 +++++++++++++---
> 1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c
> b/drivers/net/wireless/marvell/mwifiex/pcie.c
> index 4b89f557d0b6..9f9ea1350591 100644
> --- a/drivers/net/wireless/marvell/mwifiex/pcie.c
> +++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
> @@ -2050,7 +2050,7 @@ static int mwifiex_prog_fw_w_helper(struct
> mwifiex_adapter *adapter,
> }
>
> /* Wait for the command done interrupt */
> - do {
> + for (tries = 0; tries < MAX_POLL_TRIES; tries++) {
> if (mwifiex_read_reg(adapter, PCIE_CPU_INT_STATUS,
> &ireg_intr)) {
> mwifiex_dbg(adapter, ERROR,
> @@ -2062,8 +2062,18 @@ static int mwifiex_prog_fw_w_helper(struct
> mwifiex_adapter *adapter,
> ret = -1;
> goto done;
> }
> - } while ((ireg_intr & CPU_INTR_DOOR_BELL) ==
> - CPU_INTR_DOOR_BELL);
> + if (!(ireg_intr & CPU_INTR_DOOR_BELL))
> + break;
> + usleep_range(10, 20);
> + }
> + if (ireg_intr & CPU_INTR_DOOR_BELL) {
> + mwifiex_dbg(adapter, ERROR, "%s: Card failed to ACK
> download\n",
> + __func__);
> + mwifiex_unmap_pci_memory(adapter, skb,
> + PCI_DMA_TODEVICE);
> + ret = -1;
> + goto done;
> + }
>
> mwifiex_unmap_pci_memory(adapter, skb, PCI_DMA_TODEVICE);
>
Patch looks fine to me.
Acked-by: Amitkumar Karwar <[email protected]>
Regards,
Amitkumar
Brian Norris <[email protected]> wrote:
> Marvell Wifi PCIe modules don't always behave nicely for PCIe power
> management when their firmware hasn't been loaded, particularly after
> suspending the PCIe link one or more times. When this happens, we might
> end up spinning forever in this status-polling tight loop. Let's make
> this less tight by adding a timeout and by sleeping a bit in between
> reads, as we do with the other similar loops.
>
> This prevents us from hogging a CPU even in such pathological cases, and
> allows the FW initialization to just fail gracefully instead.
>
> I chose the same polling parameters as the earlier loop in this
> function, and empirically, I found that this loop never makes it more
> than about 12 cycles in a sane FW init sequence. I had no official
> information on the actual intended latency for this portion of the
> download.
>
> Signed-off-by: Brian Norris <[email protected]>
> Acked-by: Amitkumar Karwar <[email protected]>
Patch applied to wireless-drivers-next.git, thanks.
22dde1ed5a48 mwifiex: pcie: implement timeout loop for FW programming doorbell
--
https://patchwork.kernel.org/patch/9442499/
Documentation about submitting wireless patches and checking status
from patchwork:
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches