RE: [PATCH] [v2] PCI: cadence: Set LTSSM Detect Quiet state minimum delay as workaround for training defect.

From: Athani Nadeem Ladkhan
Date: Tue Jun 22 2021 - 01:13:43 EST

Next message: Ian Rogers: "Re: [PATCH 2/2] perf script: delete evlist when deleting session"
Previous message: Bernd Edlinger: "Re: [PATCH v9] exec: Fix dead-lock in de_thread with ptrace_attach"
In reply to: Lorenzo Pieralisi: "Re: [PATCH] [v2] PCI: cadence: Set LTSSM Detect Quiet state minimum delay as workaround for training defect."
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi Lorenzo,

> -----Original Message-----
> From: Lorenzo Pieralisi <lorenzo.pieralisi@xxxxxxx>
> Sent: Monday, June 21, 2021 10:48 PM
> To: Athani Nadeem Ladkhan <nadeem@xxxxxxxxxxx>
> Cc: Tom Joseph <tjoseph@xxxxxxxxxxx>; robh@xxxxxxxxxx;
> bhelgaas@xxxxxxxxxx; linux-pci@xxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; kw@xxxxxxxxx; kishon@xxxxxx; Milind Parab
> <mparab@xxxxxxxxxxx>; Swapnil Kashinath Jakhade
> <sjakhade@xxxxxxxxxxx>; Parshuram Raju Thombare
> <pthombar@xxxxxxxxxxx>
> Subject: Re: [PATCH] [v2] PCI: cadence: Set LTSSM Detect Quiet state
> minimum delay as workaround for training defect.
>
> EXTERNAL MAIL
>
>
> Subjects should not end with a period, remove it.
This will be corrected in next patch.
>
> On Fri, May 28, 2021 at 05:56:26PM +0200, Nadeem Athani wrote:
> > PCIe fails to link up if SERDES lanes not used by PCIe are assigned to
> > another protocol. For example, link training fails if lanes 2 and 3
> > are assigned to another protocol while lanes 0 and 1 are used for PCIe
> > to form a two lane link. This failure is due to an incorrect tie-off
> > on an internal status signal indicating electrical idle.
> >
> > Status signals going from SERDES to PCIe Controller are tied-off when
> > a lane is not assigned to PCIe. Signal indicating electrical idle is
> > incorrectly tied-off to a state that indicates non-idle. As a result,
> > PCIe sees unused lanes to be out of electrical idle and this causes
> > LTSSM to exit Detect.Quiet state without waiting for 12ms timeout to
> > occur. If a receiver is not detected on the first receiver detection
> > attempt in Detect.Active state, LTSSM goes back to Detect.Quiet and
> > again moves forward to Detect.Active state without waiting for 12ms as
> > required by PCIe base specification. Since wait time in Detect.Quiet
> > is skipped, multiple receiver detect operations are performed
> > back-to-back without allowing time for capacitance on the transmit
> > lines to discharge. This causes subsequent receiver detection to
> > always fail even if a receiver gets connected eventually.
> >
> > Adding a quirk flag "quirk_detect_quiet_flag" to program the minimum
> > time that LTSSM waits on entering Detect.Quiet state.
> > Setting this to 2ms for specific TI j7200 SOC as a workaround to
> > resolve a link training issue in IP.
> > In future revisions this setting will not be required.
> >
> > As per PCIe specification, all Receivers must meet the Z-RX-DC
> > specification for 2.5 GT/s within 1ms of entering Detect.Quiet LTSSM
> > substate. The LTSSM must stay in this substate until the ZRXDC
> > specification for 2.5 GT/s is met.
> >
> > 00 : 0 minimum wait time in Detect.Quiet state.
> > 01 : 100us minimum wait time in Detect.Quiet state.
> > 10 : 1ms minimum wait time in Detect.Quiet state.
> > 11 : 2ms minimum wait time in Detect.Quiet state.
> >
> > Changes in v2:
> > 1. Adding the function cdns_pcie_detect_quiet_min_delay_set in
> > pcie-cadence.c and invoking it from host and endpoint driver file.
> >
> > Signed-off-by: Nadeem Athani <nadeem@xxxxxxxxxxx>
> > ---
> > drivers/pci/controller/cadence/pcie-cadence-ep.c | 4 ++++
> > drivers/pci/controller/cadence/pcie-cadence-host.c | 3 +++
> > drivers/pci/controller/cadence/pcie-cadence.c | 17 +++++++++++++++++
> > drivers/pci/controller/cadence/pcie-cadence.h | 15 +++++++++++++++
> > 4 files changed, 39 insertions(+)
>
> Can you tell me please what's the status of these patches ?
These should be discarded.
>
> https://urldefense.com/v3/__https://patchwork.kernel.org/user/todo/linux-
> pci/?series=&submitter=194539&state=&q=&archive=__;!!EHscmS1ygiU1lA!S
> ffFfz2_4D72wnwKo_Hgq310uaxhkR7jk4l5vQMYCgwFPYLHDivDXtqJ1uu_RA$
>
> Are they all solving the same problem so I only have to review:
Only need to review the below one.
>
> https://urldefense.com/v3/__https://patchwork.kernel.org/project/linux-
> pci/patch/20210528155626.21793-1-
> nadeem@cadence.com__;!!EHscmS1ygiU1lA!SffFfz2_4D72wnwKo_Hgq310ua
> xhkR7jk4l5vQMYCgwFPYLHDivDXtorT4PhuQ$
>
> ?
>
> I am asking because the "first" posting was a two-patch series and then it
> became one, I lost track of versions in between.
>
> Thanks,
> Lorenzo
>
> > diff --git a/drivers/pci/controller/cadence/pcie-cadence-ep.c
> > b/drivers/pci/controller/cadence/pcie-cadence-ep.c
> > index 897cdde02bd8..dd7df1ac7fda 100644
> > --- a/drivers/pci/controller/cadence/pcie-cadence-ep.c
> > +++ b/drivers/pci/controller/cadence/pcie-cadence-ep.c
> > @@ -623,6 +623,10 @@ int cdns_pcie_ep_setup(struct cdns_pcie_ep *ep)
> > ep->irq_pci_addr = CDNS_PCIE_EP_IRQ_PCI_ADDR_NONE;
> > /* Reserve region 0 for IRQs */
> > set_bit(0, &ep->ob_region_map);
> > +
> > + if (ep->quirk_detect_quiet_flag)
> > + cdns_pcie_detect_quiet_min_delay_set(&ep->pcie);
> > +
> > spin_lock_init(&ep->lock);
> >
> > return 0;
> > diff --git a/drivers/pci/controller/cadence/pcie-cadence-host.c
> > b/drivers/pci/controller/cadence/pcie-cadence-host.c
> > index ae1c55503513..fb96d37a135c 100644
> > --- a/drivers/pci/controller/cadence/pcie-cadence-host.c
> > +++ b/drivers/pci/controller/cadence/pcie-cadence-host.c
> > @@ -498,6 +498,9 @@ int cdns_pcie_host_setup(struct cdns_pcie_rc *rc)
> > return PTR_ERR(rc->cfg_base);
> > rc->cfg_res = res;
> >
> > + if (rc->quirk_detect_quiet_flag)
> > + cdns_pcie_detect_quiet_min_delay_set(&rc->pcie);
> > +
> > ret = cdns_pcie_start_link(pcie);
> > if (ret) {
> > dev_err(dev, "Failed to start link\n"); diff --git
> > a/drivers/pci/controller/cadence/pcie-cadence.c
> > b/drivers/pci/controller/cadence/pcie-cadence.c
> > index 3c3646502d05..65b6c8bed0d4 100644
> > --- a/drivers/pci/controller/cadence/pcie-cadence.c
> > +++ b/drivers/pci/controller/cadence/pcie-cadence.c
> > @@ -7,6 +7,23 @@
> >
> > #include "pcie-cadence.h"
> >
> > +void cdns_pcie_detect_quiet_min_delay_set(struct cdns_pcie *pcie) {
> > + u32 delay = 0x3;
> > + u32 ltssm_control_cap;
> > +
> > + /*
> > + * Set the LTSSM Detect Quiet state min. delay to 2ms.
> > + */
> > +
> > + ltssm_control_cap = cdns_pcie_readl(pcie,
> CDNS_PCIE_LTSSM_CONTROL_CAP);
> > + ltssm_control_cap = ((ltssm_control_cap &
> > + ~CDNS_PCIE_DETECT_QUIET_MIN_DELAY_MASK) |
> > + CDNS_PCIE_DETECT_QUIET_MIN_DELAY(delay));
> > +
> > + cdns_pcie_writel(pcie, CDNS_PCIE_LTSSM_CONTROL_CAP,
> > +ltssm_control_cap); }
> > +
> > void cdns_pcie_set_outbound_region(struct cdns_pcie *pcie, u8 busnr, u8
> fn,
> > u32 r, bool is_io,
> > u64 cpu_addr, u64 pci_addr, size_t size) diff
> --git
> > a/drivers/pci/controller/cadence/pcie-cadence.h
> > b/drivers/pci/controller/cadence/pcie-cadence.h
> > index 254d2570f8c9..ccdf9cee9dde 100644
> > --- a/drivers/pci/controller/cadence/pcie-cadence.h
> > +++ b/drivers/pci/controller/cadence/pcie-cadence.h
> > @@ -189,6 +189,14 @@
> > /* AXI link down register */
> > #define CDNS_PCIE_AT_LINKDOWN (CDNS_PCIE_AT_BASE + 0x0824)
> >
> > +/* LTSSM Capabilities register */
> > +#define CDNS_PCIE_LTSSM_CONTROL_CAP (CDNS_PCIE_LM_BASE +
> 0x0054)
> > +#define CDNS_PCIE_DETECT_QUIET_MIN_DELAY_MASK GENMASK(2, 1)
> #define
> > +CDNS_PCIE_DETECT_QUIET_MIN_DELAY_SHIFT 1 #define
> > +CDNS_PCIE_DETECT_QUIET_MIN_DELAY(delay) \
> > + (((delay) << CDNS_PCIE_DETECT_QUIET_MIN_DELAY_SHIFT) & \
> > + CDNS_PCIE_DETECT_QUIET_MIN_DELAY_MASK)
> > +
> > enum cdns_pcie_rp_bar {
> > RP_BAR_UNDEFINED = -1,
> > RP_BAR0,
> > @@ -292,6 +300,7 @@ struct cdns_pcie {
> > * @avail_ib_bar: Satus of RP_BAR0, RP_BAR1 and RP_NO_BAR if it's free
> or
> > * available
> > * @quirk_retrain_flag: Retrain link as quirk for PCIe Gen2
> > + * @quirk_detect_quiet_flag: LTSSM Detect Quiet min delay set as
> > + quirk
> > */
> > struct cdns_pcie_rc {
> > struct cdns_pcie pcie;
> > @@ -301,6 +310,7 @@ struct cdns_pcie_rc {
> > u32 device_id;
> > bool avail_ib_bar[CDNS_PCIE_RP_MAX_IB];
> > bool quirk_retrain_flag;
> > + bool quirk_detect_quiet_flag;
> > };
> >
> > /**
> > @@ -331,6 +341,7 @@ struct cdns_pcie_epf {
> > * registers fields (RMW) accessible by both remote RC and EP to
> > * minimize time between read and write
> > * @epf: Structure to hold info about endpoint function
> > + * @quirk_detect_quiet_flag: LTSSM Detect Quiet min delay set as
> > + quirk
> > */
> > struct cdns_pcie_ep {
> > struct cdns_pcie pcie;
> > @@ -345,6 +356,7 @@ struct cdns_pcie_ep {
> > /* protect writing to PCI_STATUS while raising legacy interrupts */
> > spinlock_t lock;
> > struct cdns_pcie_epf *epf;
> > + bool quirk_detect_quiet_flag;
> > };
> >
> >
> > @@ -505,6 +517,9 @@ static inline int cdns_pcie_ep_setup(struct
> cdns_pcie_ep *ep)
> > return 0;
> > }
> > #endif
> > +
> > +void cdns_pcie_detect_quiet_min_delay_set(struct cdns_pcie *pcie);
> > +
> > void cdns_pcie_set_outbound_region(struct cdns_pcie *pcie, u8 busnr, u8
> fn,
> > u32 r, bool is_io,
> > u64 cpu_addr, u64 pci_addr, size_t size);
> > --
> > 2.15.0
> >

Next message: Ian Rogers: "Re: [PATCH 2/2] perf script: delete evlist when deleting session"
Previous message: Bernd Edlinger: "Re: [PATCH v9] exec: Fix dead-lock in de_thread with ptrace_attach"
In reply to: Lorenzo Pieralisi: "Re: [PATCH] [v2] PCI: cadence: Set LTSSM Detect Quiet state minimum delay as workaround for training defect."
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]