usb: host: add XHCI_CDNS_HOST flag
Subsystems (2)
PATCH 1/2 usb: host: add XHCI_CDNS_HOST flag
The Cadence xHCI host has the same issue with Intel's, it is triggered by reboot stress test.
--- drivers/usb/host/xhci.c | 2 +- drivers/usb/host/xhci.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)
drivers/usb/host/xhci.c +1 -1
@@ -193,7 +193,7 @@ int xhci_reset(struct xhci_hcd *xhci)* Without this delay, the subsequent HC register access, * may result in a system hang very rarely. */ -if (xhci->quirks &XHCI_INTEL_HOST ) +if (xhci->quirks & (XHCI_INTEL_HOST |XHCI_CDNS_HOST ))udelay (1000 ); ret = xhci_handshake(&xhci->op_regs->command,
drivers/usb/host/xhci.h +1 -0
@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_CDNS_HOST BIT_ULL(38)unsigned int num_active_eps;unsigned int limit_active_eps;
Signature · 1 line
2.17.1
PATCH 2/2 usb: host: xhci-plat: forbid runtime pm conditionally
Some users may want to enable runtime pm by default, using the quirk to judge it.
--- drivers/usb/host/xhci-plat.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/usb/host/xhci-plat.c +2 -1
@@ -351,7 +351,8 @@ static int xhci_plat_probe(struct platform_device *pdev)* Prevent runtime pm from being on as default, users should enable * runtime pm using power/control in sysfs. */ -pm_runtime_forbid (&pdev->dev); +if (!(xhci->quirks &XHCI_DEFAULT_PM_RUNTIME_ALLOW )) +pm_runtime_forbid (&pdev->dev);return 0 ;
Signature · 1 line
2.17.1
Hi Peter,
Quoted 1 line · Peter Chen
The Cadence xHCI host has the same issue with Intel's,
s/with/as
Quoted 1 line
it is triggered by reboot stress test.
Can you please provide some more details about the test so I can try at my end. Thanks.
cheers,
-roger
Quoted 35 lines
Cc Pawel Laszczak <[email protected]>Cc Roger Quadros <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/host/xhci.c | 2 +- drivers/usb/host/xhci.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)drivers/usb/host/xhci.c +1
-1@@ -193,7 +193,7 @@ int xhci_reset(struct xhci_hcd *xhci)* Without this delay, the subsequent HC register access, * may result in a system hang very rarely. */ -if (xhci->quirks &XHCI_INTEL_HOST ) +if (xhci->quirks & (XHCI_INTEL_HOST |XHCI_CDNS_HOST ))udelay (1000 ); ret = xhci_handshake(&xhci->op_regs->command,drivers/usb/host/xhci.h +1
-0@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_CDNS_HOST BIT_ULL(38)unsigned int num_active_eps;unsigned int limit_active_eps;
Signature · 2 lines
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki. Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki
Quoted 12 lines · Peter Chen
Quoted 1 line
The Cadence xHCI host has the same issue with Intel's,
s/with/as
Quoted 1 line
it is triggered by reboot stress test.
Can you please provide some more details about the test so I can try at my end. Thanks.
Connect USB3 Drive at port, and reboot system over night.
Peter
Quoted 30 lines · Peter Chen
The Cadence xHCI host has the same issue with Intel's, it is triggered by reboot stress test.
Cc Pawel Laszczak <[email protected]>Cc Roger Quadros <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/host/xhci.c | 2 +- drivers/usb/host/xhci.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)drivers/usb/host/xhci.c +1
-1@@ -193,7 +193,7 @@ int xhci_reset(struct xhci_hcd *xhci)* Without this delay, the subsequent HC register access, * may result in a system hang very rarely. */ -if (xhci->quirks &XHCI_INTEL_HOST ) +if (xhci->quirks & (XHCI_INTEL_HOST |XHCI_CDNS_HOST ))udelay (1000 ); ret = xhci_handshake(&xhci->op_regs->command,drivers/usb/host/xhci.h +0
-0
drivers/usb/host/xhci.h struct xhci_hcd { · line 1883
Quoted 7 lines
drivers/usb/host/xhci.h +1
-0@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_CDNS_HOST BIT_ULL(38)unsigned int num_active_eps;unsigned int limit_active_eps;
Is the XHCI_CDNS_HOST quirk bit set in some other patchseries?
-Mathias
Quoted 32 lines
Quoted 32 lines
Cc Pawel Laszczak <[email protected]>Cc Roger Quadros <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/host/xhci.c | 2 +- drivers/usb/host/xhci.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)drivers/usb/host/xhci.c index renamed from drivers/usb/host/xhci.c +0
-0
482fe8c5e3b4..fc72a03dc27f100644drivers/usb/host/xhci.c +1
-1@@ -193,7 +193,7 @@ int xhci_reset(struct xhci_hcd *xhci)* Without this delay, the subsequent HC register access, * may result in a system hang very rarely. */ -if (xhci->quirks &XHCI_INTEL_HOST ) +if (xhci->quirks & (XHCI_INTEL_HOST |XHCI_CDNS_HOST ))udelay (1000 ); ret = xhci_handshake(&xhci->op_regs->command,drivers/usb/host/xhci.h index renamed from drivers/usb/host/xhci.h +0
-0
8be88379c0fb..4b7275c73ea5100644drivers/usb/host/xhci.h +1
-0@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_CDNS_HOST BIT_ULL(38)
Quoted 6 lines
drivers/usb/host/xhci.h struct xhci_hcd { · line 1883
Quoted 3 lines
drivers/usb/host/xhci.h +0
-0unsigned int num_active_eps;unsigned int limit_active_eps;Is the
XHCI_CDNS_HOSTquirk bit set in some other patchseries?
Currently, no, only at the function of xhci_reset in this commit.
Peter
Quoted 32 lines · Peter Chen
Quoted 32 lines
Quoted 32 lines
Cc Pawel Laszczak <[email protected]>Cc Roger Quadros <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/host/xhci.c | 2 +- drivers/usb/host/xhci.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)drivers/usb/host/xhci.c index renamed from drivers/usb/host/xhci.c +0
-0
482fe8c5e3b4..fc72a03dc27f100644drivers/usb/host/xhci.c +1
-1@@ -193,7 +193,7 @@ int xhci_reset(struct xhci_hcd *xhci)* Without this delay, the subsequent HC register access, * may result in a system hang very rarely. */ -if (xhci->quirks &XHCI_INTEL_HOST ) +if (xhci->quirks & (XHCI_INTEL_HOST |XHCI_CDNS_HOST ))udelay (1000 ); ret = xhci_handshake(&xhci->op_regs->command,drivers/usb/host/xhci.h index renamed from drivers/usb/host/xhci.h +0
-0
8be88379c0fb..4b7275c73ea5100644drivers/usb/host/xhci.h +1
-0@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_CDNS_HOST BIT_ULL(38)
Quoted 8 lines
Quoted 6 lines
drivers/usb/host/xhci.h struct xhci_hcd { · line 1883
Quoted 3 lines
drivers/usb/host/xhci.h +0
-0unsigned int num_active_eps;unsigned int limit_active_eps;Is the
XHCI_CDNS_HOSTquirk bit set in some other patchseries?Currently, no, only at the function of
xhci_resetin this commit.
I can only see it being checked, not set. Am I missing something?
-Mathias
Quoted 40 lines · Mathias Nyman
Quoted 39 lines · Peter Chen
Quoted 38 lines
Quoted 28 lines
Cc Pawel Laszczak <[email protected]>Cc Roger Quadros <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/host/xhci.c | 2 +- drivers/usb/host/xhci.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)drivers/usb/host/xhci.c index renamed from drivers/usb/host/xhci.c +0
-0
482fe8c5e3b4..fc72a03dc27f100644drivers/usb/host/xhci.c +1
-1@@ -193,7 +193,7 @@ int xhci_reset(struct xhci_hcd *xhci)* Without this delay, the subsequent HC register access, * may result in a system hang very rarely. */ -if (xhci->quirks &XHCI_INTEL_HOST ) +if (xhci->quirks & (XHCI_INTEL_HOST |XHCI_CDNS_HOST ))udelay (1000 ); ret = xhci_handshake(&xhci->op_regs->command,drivers/usb/host/xhci.h index renamed from drivers/usb/host/xhci.h +0
-0
8be88379c0fb..4b7275c73ea5100644drivers/usb/host/xhci.h struct xhci_hcd { · line 1883
Quoted 7 lines
drivers/usb/host/xhci.h +1
-0@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_CDNS_HOST BIT_ULL(38)unsigned int num_active_eps;unsigned int limit_active_eps;Is the
XHCI_CDNS_HOSTquirk bit set in some other patchseries?
Quoted 6 lines
Quoted 1 line
Currently, no, only at the function of
xhci_resetin this commit.I can only see it being checked, not set. Am I missing something?
-Mathias
Oh, there is a missing patch for cdns3 host change, the code is located at drivers/usb/cdns3/host.c. I will merge them together, and send v2.
Signature · 2 lines
Thanks, Peter Chen
Quoted 11 lines
Quoted 11 lines
Quoted 10 lines
Quoted 8 lines
drivers/usb/host/xhci.h struct xhci_hcd { · line 1883
Quoted 5 lines
drivers/usb/host/xhci.h +0
-0unsigned int num_active_eps;unsigned int limit_active_eps;Is the
XHCI_CDNS_HOSTquirk bit set in some other patchseries?Currently, no, only at the function of
xhci_resetin this commit.
Quoted 8 lines
Quoted 4 lines
I can only see it being checked, not set. Am I missing something?
-Mathias
Oh, there is a missing patch for cdns3 host change, the code is located at
drivers/usb/cdns3/host.c. I will merge them together, and send v2.
Mathias, adding XHCI_CDNS_HOST quirk at cdns3 part depends on some reviewing patches [1], these patches add the struct xhci_plat_priv with other quirks first. Would you help queue these cdns3 host patches after reviewing? Or what's the suggestion to handle this dependency? Thanks.
[1] https://patchwork.kernel.org/project/linux-usb/patch/[email protected]/
Peter
Quoted 6 lines
Quoted 5 lines
Oh, there is a missing patch for cdns3 host change, the code is located at
drivers/usb/cdns3/host.c. I will merge them together, and send v2.
Quoted 9 lines
Mathias, adding
XHCI_CDNS_HOSTquirk at cdns3 part depends on some reviewing patches [1], these patches add the structxhci_plat_privwith other quirks first. Would you help queue these cdns3 host patches after reviewing? Or what's the suggestion to handle this dependency? Thanks.[1]
https://patchwork.kernel.org/project/linux-usb/patch/20201022013930.2166-1
[email protected]/
I think we may not need this patch at upstream kernel, I run the test overnight, and could not reproduce hang issue. @Roger Quadros, did you meet hang at the reboot stress test?
Peter