| Message ID | 20220508070905.1878172-1-sasha.neftin@intel.com |
|---|---|
| State | Accepted |
| Delegated to: | Anthony Nguyen |
| Headers | show |
| Series | [v2,1/1] e1000e: Enable GPT clock before sending message to CSME | expand |
Dear Sasha, Am 08.05.22 um 09:09 schrieb Sasha Neftin: > On corporate (CSME) ADL systems, the Ethernet Controller may stop working > ("HW unit hang") after exiting from the s0ix state. The reason is that > CSME misses the message sent by the host. What message exactly by the way? > Enabling the dynamic GPT clock solves this problem. This clock is > cleared upon HW initialization. “Enabling” sounds like it’s not running before. But reading the comment, it’s running. Could you make that more clear please? The GPT clock is running when in S0ix? Lastly, is that a CSME firmware bug and what CSME firmware version was it? > Fixes: 3e55d231716e ("e1000e: Add handshake with the CSME to support S0ix") > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214821 > Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com> > Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> > --- > v2: > Correct commit message and minor fix in comment. > drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c > index fa06f68c8c80..c64102b29862 100644 > --- a/drivers/net/ethernet/intel/e1000e/netdev.c > +++ b/drivers/net/ethernet/intel/e1000e/netdev.c > @@ -6494,6 +6494,10 @@ static void e1000e_s0ix_exit_flow(struct e1000_adapter *adapter) > > if (er32(FWSM) & E1000_ICH_FWSM_FW_VALID && > hw->mac.type >= e1000_pch_adp) { > + /* Keep the GPT clock enabled for CSME */ > + mac_data = er32(FEXTNVM); > + mac_data |= BIT(3); > + ew32(FEXTNVM, mac_data); > /* Request ME unconfigure the device from S0ix */ > mac_data = er32(H2ME); > mac_data &= ~E1000_H2ME_START_DPG; Kind regards, Paul
On Sun, May 08, 2022 at 10:09:05AM +0300, Sasha Neftin wrote: > On corporate (CSME) ADL systems, the Ethernet Controller may stop working > ("HW unit hang") after exiting from the s0ix state. The reason is that > CSME misses the message sent by the host. Enabling the dynamic GPT clock > solves this problem. This clock is cleared upon HW initialization. > > Fixes: 3e55d231716e ("e1000e: Add handshake with the CSME to support S0ix") > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214821 > Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com> > Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> This commit fixed the hardware unit hang issue on my machine. Tested-by: Chia-Lin Kao (AceLan) <acelan.kao@canonical.com> > --- > v2: > Correct commit message and minor fix in comment. > drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c > index fa06f68c8c80..c64102b29862 100644 > --- a/drivers/net/ethernet/intel/e1000e/netdev.c > +++ b/drivers/net/ethernet/intel/e1000e/netdev.c > @@ -6494,6 +6494,10 @@ static void e1000e_s0ix_exit_flow(struct e1000_adapter *adapter) > > if (er32(FWSM) & E1000_ICH_FWSM_FW_VALID && > hw->mac.type >= e1000_pch_adp) { > + /* Keep the GPT clock enabled for CSME */ > + mac_data = er32(FEXTNVM); > + mac_data |= BIT(3); > + ew32(FEXTNVM, mac_data); > /* Request ME unconfigure the device from S0ix */ > mac_data = er32(H2ME); > mac_data &= ~E1000_H2ME_START_DPG;
On 5/8/2022 10:09, Sasha Neftin wrote: > On corporate (CSME) ADL systems, the Ethernet Controller may stop working > ("HW unit hang") after exiting from the s0ix state. The reason is that > CSME misses the message sent by the host. Enabling the dynamic GPT clock > solves this problem. This clock is cleared upon HW initialization. > > Fixes: 3e55d231716e ("e1000e: Add handshake with the CSME to support S0ix") > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214821 > Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com> > Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> > --- > v2: > Correct commit message and minor fix in comment. > drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++++ > 1 file changed, 4 insertions(+) Tested-by: Naama Meir <naamax.meir@linux.intel.com>
diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c index fa06f68c8c80..c64102b29862 100644 --- a/drivers/net/ethernet/intel/e1000e/netdev.c +++ b/drivers/net/ethernet/intel/e1000e/netdev.c @@ -6494,6 +6494,10 @@ static void e1000e_s0ix_exit_flow(struct e1000_adapter *adapter) if (er32(FWSM) & E1000_ICH_FWSM_FW_VALID && hw->mac.type >= e1000_pch_adp) { + /* Keep the GPT clock enabled for CSME */ + mac_data = er32(FEXTNVM); + mac_data |= BIT(3); + ew32(FEXTNVM, mac_data); /* Request ME unconfigure the device from S0ix */ mac_data = er32(H2ME); mac_data &= ~E1000_H2ME_START_DPG;
On corporate (CSME) ADL systems, the Ethernet Controller may stop working ("HW unit hang") after exiting from the s0ix state. The reason is that CSME misses the message sent by the host. Enabling the dynamic GPT clock solves this problem. This clock is cleared upon HW initialization. Fixes: 3e55d231716e ("e1000e: Add handshake with the CSME to support S0ix") Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214821 Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com> Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> --- v2: Correct commit message and minor fix in comment. drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++++ 1 file changed, 4 insertions(+)