ptp: Introduce strict checking of external time stamp options.
Subsystems (9)
Richard Cochran
maintainerPATCH 8/13 ptp: Introduce strict checking of external time stamp options.
User space may request time stamps on rising edges, falling edges, or both. However, the particular mode may or may not be supported in the hardware or in the driver. This patch adds a "strict" flag that tells drivers to ensure that the requested mode will be honored.
--- drivers/net/dsa/mv88e6xxx/ptp.c | 3 ++- drivers/net/ethernet/intel/igb/igb_ptp.c | 3 ++- drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c | 3 ++- drivers/net/ethernet/renesas/ravb_ptp.c | 3 ++- drivers/net/phy/dp83640.c | 3 ++- drivers/ptp/ptp_chardev.c | 2 ++ include/uapi/linux/ptp_clock.h | 4 +++- 7 files changed, 15 insertions(+), 6 deletions(-)
drivers/net/dsa/mv88e6xxx/ptp.c +2 -1
@@ -276,7 +276,8 @@ static int mv88e6352_ptp_enable_extts(struct mv88e6xxx_chip *chip,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ; pin =ptp_find_pin (chip->ptp_clock,PTP_PF_EXTTS , rq->extts.index);
drivers/net/ethernet/intel/igb/igb_ptp.c +2 -1
@@ -524,7 +524,8 @@ static int igb_ptp_feature_enable_i210(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ;if (on) {
drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +2 -1
@@ -239,7 +239,8 @@ static int mlx5_extts_configure(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ;if (rq->extts.index >= clock->ptp_info.n_pins)
drivers/net/ethernet/renesas/ravb_ptp.c +2 -1
@@ -185,7 +185,8 @@ static int ravb_ptp_extts(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (req->flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ;if (req->index)
drivers/net/phy/dp83640.c +2 -1
@@ -472,7 +472,8 @@ static int ptp_dp83640_enable(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ; index = rq->extts.index;if (index >=N_EXT_TS )
drivers/ptp/ptp_chardev.c +2 -0
@@ -150,6 +150,8 @@ long ptp_ioctl(struct posix_clock *pc, unsigned int cmd, unsigned long arg)break ; }if (cmd ==PTP_EXTTS_REQUEST2 ) { +/* Tell the drivers to check the flags carefully. */ + req.extts.flags |=PTP_STRICT_FLAGS ;/* Make sure no reserved bit is set. */ if ((req.extts.flags & ~PTP_EXTTS_VALID_FLAGS ) || req.extts.rsv[0 ] || req.extts.rsv[1 ]) {
include/uapi/linux/ptp_clock.h +3 -1
@@ -31,6 +31,7 @@#define PTP_ENABLE_FEATURE (1<<0)#define PTP_RISING_EDGE (1<<1)#define PTP_FALLING_EDGE (1<<2) +#define PTP_STRICT_FLAGS (1<<3)#define PTP_EXTTS_EDGES (PTP_RISING_EDGE | PTP_FALLING_EDGE)/*
@@ -38,7 +39,8 @@*/ #define PTP_EXTTS_VALID_FLAGS (PTP_ENABLE_FEATURE | \ PTP_RISING_EDGE | \ - PTP_FALLING_EDGE) + PTP_FALLING_EDGE | \ + PTP_STRICT_FLAGS)/* * flag fields valid for the original PTP_EXTTS_REQUEST ioctl.
Signature · 1 line
2.20.1
Quoted 15 lines
-----Original Message-----
From: Richard Cochran <[email protected]>
Sent: Thursday, November 14, 2019 10:45 AM
To: netdev at vger.kernel.org
Cc: intel-wired-lan at lists.osuosl.org; David Miller <[email protected]>; Brandon Streiff <[email protected]>; Hall, Christopher S <[email protected]>; Eugenia Emantayev <[email protected]>; Felipe Balbi <[email protected]>; Feras Daoud <[email protected]>; Keller, Jacob E <[email protected]>; Kirsher, Jeffrey T <[email protected]>; Sergei Shtylyov <[email protected]>; Stefan Sorensen <[email protected]>
Subject: [PATCH net 08/13] ptp: Introduce strict checking of external time stamp options.
Quoted 5 lines
User space may request time stamps on rising edges, falling edges, or both. However, the particular mode may or may not be supported in the hardware or in the driver. This patch adds a "strict" flag that tells drivers to ensure that the requested mode will be honored.
So, this patch adds the flag *and* modifies the drivers to accept it, but not actually enable strict checking?
I'd prefer if this flag got added, and the drivers were modified in separate patches to both allow the flag and to perform the strict check.. that feels like a cleaner patch boundary.
That would ofcourse break the drivers that reject the strict command until they're fixed in follow-on commands.. hmm
Thanks,
Jake
Quoted 126 lines
Signed-off-by Richard Cochran <[email protected]>--- drivers/net/dsa/mv88e6xxx/ptp.c | 3 ++- drivers/net/ethernet/intel/igb/igb_ptp.c | 3 ++- drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c | 3 ++- drivers/net/ethernet/renesas/ravb_ptp.c | 3 ++- drivers/net/phy/dp83640.c | 3 ++- drivers/ptp/ptp_chardev.c | 2 ++ include/uapi/linux/ptp_clock.h | 4 +++- 7 files changed, 15 insertions(+), 6 deletions(-)drivers/net/dsa/mv88e6xxx/ptp.c +2
-1@@ -276,7 +276,8 @@ static int mv88e6352_ptp_enable_extts(struct mv88e6xxx_chip *chip,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS )) return -EOPNOTSUPP;drivers/net/dsa/mv88e6xxx/ptp.c +0
-0pin =ptp_find_pin (chip->ptp_clock,PTP_PF_EXTTS , rq->extts.index);drivers/net/ethernet/intel/igb/igb_ptp.c +2
-1@@ -524,7 +524,8 @@ static int igb_ptp_feature_enable_i210(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS )) return -EOPNOTSUPP;drivers/net/ethernet/intel/igb/igb_ptp.c +0
-0if (on) {drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +2
-1@@ -239,7 +239,8 @@ static int mlx5_extts_configure(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS )) return -EOPNOTSUPP;drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +0
-0if (rq->extts.index >= clock->ptp_info.n_pins)drivers/net/ethernet/renesas/ravb_ptp.c +2
-1@@ -185,7 +185,8 @@ static int ravb_ptp_extts(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (req->flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ;if (req->index)drivers/net/phy/dp83640.c +2
-1@@ -472,7 +472,8 @@ static int ptp_dp83640_enable(struct ptp_clock_info *ptp,/* Reject requests with unsupported flags */ if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |PTP_RISING_EDGE | -PTP_FALLING_EDGE )) +PTP_FALLING_EDGE | +PTP_STRICT_FLAGS ))return -EOPNOTSUPP ; index = rq->extts.index;if (index >=N_EXT_TS )drivers/ptp/ptp_chardev.c +2
-0@@ -150,6 +150,8 @@ long ptp_ioctl(struct posix_clock *pc, unsigned int cmd, unsigned long arg) break; }if (cmd ==PTP_EXTTS_REQUEST2 ) { +/* Tell the drivers to check the flags carefully. */ + req.extts.flags |=PTP_STRICT_FLAGS ;/* Make sure no reserved bit is set. */ if ((req.extts.flags & ~PTP_EXTTS_VALID_FLAGS) ||drivers/ptp/ptp_chardev.c +0
-0req.extts.rsv[0] || req.extts.rsv[1]) {include/uapi/linux/ptp_clock.h +3
-2@@ -31,6 +31,7 @@#define PTP_ENABLE_FEATURE (1<<0)#define PTP_RISING_EDGE (1<<1)#define PTP_FALLING_EDGE (1<<2) +#define PTP_STRICT_FLAGS (1<<3)#define PTP_EXTTS_EDGES (PTP_RISING_EDGE | PTP_FALLING_EDGE)/* @@ -38,7 +39,8 @@*/ #define PTP_EXTTS_VALID_FLAGS (PTP_ENABLE_FEATURE | \ PTP_RISING_EDGE | \ - PTP_FALLING_EDGE) + PTP_FALLING_EDGE | \ + PTP_STRICT_FLAGS)/* * flag fields valid for the original PTP_EXTTS_REQUEST ioctl. -- 2.20.1
Richard Cochran
maintainerQuoted 5 lines · Jacob E
So, this patch adds the flag *and* modifies the drivers to accept it, but not actually enable strict checking?
I'd prefer if this flag got added, and the drivers were modified in separate patches to both allow the flag and to perform the strict check.. that feels like a cleaner patch boundary.
That would ofcourse break the drivers that reject the strict command until they're fixed in follow-on commands.. hmm
You are right, but if anything I'd squash the following four driver patches into this one. I left the series in little steps just to make review easier. Strictly speaking, if you were to do a git bisect from the introduction of the "2" ioctls until here, you would find drivers' acceptance of the new flags changing. But it is too late to fix that, and I doubt anyone will care.
IMHO it *is* important to have v5.4 with strict checking.
Thanks,
Richard
Quoted 33 lines
-----Original Message-----
From: Richard Cochran <[email protected]>
Sent: Thursday, November 14, 2019 11:44 AM
To: Keller, Jacob E <[email protected]>
Cc: netdev at vger.kernel.org; intel-wired-lan at lists.osuosl.org; David Miller <[email protected]>; Brandon Streiff <[email protected]>; Hall, Christopher S <[email protected]>; Eugenia Emantayev <[email protected]>; Felipe Balbi <[email protected]>; Feras Daoud <[email protected]>; Kirsher, Jeffrey T <[email protected]>; Sergei Shtylyov <[email protected]>; Stefan Sorensen <[email protected]>
Subject: Re: [PATCH net 08/13] ptp: Introduce strict checking of external time stamp options.Quoted 1 line · Jacob E
So, this patch adds the flag *and* modifies the drivers to accept it, but not
actually enable strict checking?
Quoted 2 lines
I'd prefer if this flag got added, and the drivers were modified in separate
patches to both allow the flag and to perform the strict check.. that feels like a cleaner patch boundary.
Quoted 2 lines
That would ofcourse break the drivers that reject the strict command until
they're fixed in follow-on commands.. hmm
You are right, but if anything I'd squash the following four driver patches into this one. I left the series in little steps just to make review easier. Strictly speaking, if you were to do a git bisect from the introduction of the "2" ioctls until here, you would find drivers' acceptance of the new flags changing. But it is too late to fix that, and I doubt anyone will care.
Quoted 4 lines
IMHO it *is* important to have v5.4 with strict checking.
Thanks,
Richard
Yes I agree. I think the series is good as is, and having this fixed before the ioctls have been in a full release makes sense.
Thanks,
Jake