| Message ID | 559E3098.9070609@gtsys.com.hk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi, On Thu, Jul 09, 2015 at 04:28:08PM +0800, Chris Ruehl wrote: > Uwe, > > On Thursday, July 09, 2015 03:12 PM, Uwe Kleine-König wrote: > >Hello Markus, > > > >Cc += Chris Ruehl > > > >On Wed, Jul 08, 2015 at 04:16:06PM +0200, Markus Pargmann wrote: > >>The pins do not have a 1:1 mapping from index to pin_id. Unfortunately > >>the debug output assumes exactly that. > >> > >>The first driver using imx1-core was imx27, which had exactly this 1:1 > >>mapping. It was accidently removed when removing all unused pads which > >>were listed: > >> 607af165c047 (pinctrl: i.MX27: Remove nonexistent pad definitions) > >> > >>The patch fixes this issue by printing the pin_id directly and not the > >>pin name. > >Knowing a bit about the imx pinctrl drivers I failed to understand what > >you wrote here. Probably because I first though that "1:1 mapping" is a > >hardware property. What about: > > > > Subject: pinctrl: imx1-core: Fix debug output in .pin_config_set callback > > > > imx1_pinconf_set assumes that the array of pins in struct > > imx1_pinctrl_soc_info can be indexed by pin id to get the > > pinctrl_pin_desc for a pin. This used to be correct up to commit > > 607af165c047 ("pinctrl: i.MX27: Remove nonexistent pad definitions") > > which removed some entries from the array and so made it wrong to access > > the array by pin id. > > > > Implement the easiest fix by not resolving the pin id to a name but > > printing the id instead. > > > > Fixes: 607af165c047 ("pinctrl: i.MX27: Remove nonexistent pad definitions") > > Cc: stable@vger.kernel.org > > Reported-by: Chris Ruehl <chris.ruehl@gtsys.com.hk> > > Signed-off-by: Markus Pargmann <mpa@pengutronix.de> > > > >Having the pad name in the output is nice, is it worth to search for the > >right pinctrl_pin_desc in the array? The array is still sorted, so a > >binary search would do, maybe a function for this already exists? > > > >How did Chris notice the error? Just a bogus output, or did it crash the > >kernel? That would be worth to note in the commit log, too. > > > >Otherwise the change looks fine. > > > >Best regards > >Uwe > > I had a crash on a array overrun. > > I'd applied a patch to the linux-usb (which was spit out by the maintainer > perl script) Yes, as there is already a patch from Chris my patch can be dropped. We should continue the discussion on your patch instead. > chris@wheezyvm:~/kernel.d/linux-next$ cat > 0001-Pinctrl-imx1-fix-wrong-pin-name-resolving.patch > From 50d56e5f626b2ea86211818cf487514c96f60487 Mon Sep 17 00:00:00 2001 > From: Chris Ruehl <chris.ruehl@gtsys.com.hk> > Date: Sat, 23 May 2015 15:02:44 +0800 > Subject: [PATCH] Pinctrl: imx1: fix wrong pin-name resolving > > Bug in function imx1_pinconf_set() cause crash when > princtrl debug is enabled and the pin_id becomes larger > then the info->pins[] contains. > > imx27-pinctrl 10015000.iomuxc: request pin 134 (MX27_PAD_UART2_TXD) for > 1000b000.serial > imx27-pinctrl 10015000.iomuxc: request pin 135 (MX27_PAD_UART2_RXD) for > 1000b000.serial > imx27-pinctrl 10015000.iomuxc: request pin 131 (MX27_PAD_UART2_CTS) for > 1000b000.serial > imx27-pinctrl 10015000.iomuxc: request pin 132 (MX27_PAD_UART2_RTS) for > 1000b000.serial > imx27-pinctrl 10015000.iomuxc: enable function uart group uart2-1 > imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x86, function 0, gpio 0, > direction 1, oconf 0, iconfa 0, iconfb 0 > imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x87, function 0, gpio 0, > direction 0, oconf 0, iconfa 0, iconfb 0 > imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x83, function 0, gpio 0, > direction 1, oconf 0, iconfa 0, iconfb 0 > imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x84, function 0, gpio 0, > direction 0, oconf 0, iconfa 0, iconfb 0 > imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=134 > imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_UART1_RTS > imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=135 > imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_RTCK > imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=131 > imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_UART1_TXD > imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=132 > imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_UART1_RXD > ... > imx27-pinctrl 10015000.iomuxc: write: register 0xf4415508 offset 4 value 0x3 > imx27-pinctrl 10015000.iomuxc: write: register 0xf4415510 offset 4 value 0x0 > imx27-pinctrl 10015000.iomuxc: write: register 0xf4415518 offset 4 value 0x0 > imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0xb5, function 0, gpio 0, > direction 1, oconf 0, iconfa 0, iconfb 0 > imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=171 > Unable to handle kernel paging request at virtual address 6c61765f > pgd = c0004000 > 6c61765f] *pgd=00000000 > Internal error: Oops: 5 [#1] ARM > Modules linked in: > CPU: 0 PID: 1 Comm: swapper Not tainted 4.1.0-rc4-next-20150522-dirty #8 > Hardware name: GTSYS i.MX27GTSIR (Device Tree Support) > task: ce832000 ti: ce848000 task.ti: ce848000 > PC is at strnlen+0x28/0x3c > LR is at string.isra.4+0x34/0xcc > pc : [<c01ae188>] lr : [<c01af9a4>] psr: 20000093 > sp : ce849a88 ip : ce849a98 fp : ce849a94 > r10: c05d7e3a r9 : c05d81e4 r8 : 00000000 > r7 : 6c61765f r6 : c05d81e4 r5 : ffffffff r4 : c05d7e3a > r3 : 6c61765f r2 : 6c61765f r1 : 6c61765e r0 : 6c61765f > Flags: nzCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel > > > Signed-off-by: Chris Ruehl <chris.ruehl@gtsys.com.hk> > --- > drivers/pinctrl/freescale/pinctrl-imx1-core.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/drivers/pinctrl/freescale/pinctrl-imx1-core.c > b/drivers/pinctrl/freescale/pinctrl-imx1-core.c > index 5ac59fb..8408bd8 100644 > --- a/drivers/pinctrl/freescale/pinctrl-imx1-core.c > +++ b/drivers/pinctrl/freescale/pinctrl-imx1-core.c > @@ -403,14 +403,16 @@ static int imx1_pinconf_set(struct pinctrl_dev *pctldev, > unsigned num_configs) > { > struct imx1_pinctrl *ipctl = pinctrl_dev_get_drvdata(pctldev); > - const struct imx1_pinctrl_soc_info *info = ipctl->info; > + struct pin_desc *desc; > int i; > > + desc = pin_desc_get(pctldev, pin_id); For the rest of this function pin_desc_get() is not necessary. This is not a simple function call but a radix tree lookup as far as I can see. I am not sure how important this debugging information is and how expensive this function call is?! Best Regards, Markus
Hi, On Thu, Jul 09, 2015 at 03:30:36PM +0200, Markus Pargmann wrote: > > b/drivers/pinctrl/freescale/pinctrl-imx1-core.c > > index 5ac59fb..8408bd8 100644 > > --- a/drivers/pinctrl/freescale/pinctrl-imx1-core.c > > +++ b/drivers/pinctrl/freescale/pinctrl-imx1-core.c > > @@ -403,14 +403,16 @@ static int imx1_pinconf_set(struct pinctrl_dev *pctldev, > > unsigned num_configs) > > { > > struct imx1_pinctrl *ipctl = pinctrl_dev_get_drvdata(pctldev); > > - const struct imx1_pinctrl_soc_info *info = ipctl->info; > > + struct pin_desc *desc; > > int i; > > > > + desc = pin_desc_get(pctldev, pin_id); > > For the rest of this function pin_desc_get() is not necessary. This is > not a simple function call but a radix tree lookup as far as I can see. > I am not sure how important this debugging information is and how > expensive this function call is?! If it's moved into the call to pr_debug it should be optimized away when debugging is off. That together with a merge of Chris' and mine commit log and I'm happy. Best regards Uwe
On Thursday, July 09, 2015 09:30 PM, Markus Pargmann wrote: > Hi, > > On Thu, Jul 09, 2015 at 04:28:08PM +0800, Chris Ruehl wrote: >> Uwe, >> >> On Thursday, July 09, 2015 03:12 PM, Uwe Kleine-König wrote: >>> Hello Markus, >>> >>> Cc += Chris Ruehl >>> >>> On Wed, Jul 08, 2015 at 04:16:06PM +0200, Markus Pargmann wrote: >>>> The pins do not have a 1:1 mapping from index to pin_id. Unfortunately >>>> the debug output assumes exactly that. >>>> >>>> The first driver using imx1-core was imx27, which had exactly this 1:1 >>>> mapping. It was accidently removed when removing all unused pads which >>>> were listed: >>>> 607af165c047 (pinctrl: i.MX27: Remove nonexistent pad definitions) >>>> >>>> The patch fixes this issue by printing the pin_id directly and not the >>>> pin name. >>> Knowing a bit about the imx pinctrl drivers I failed to understand what >>> you wrote here. Probably because I first though that "1:1 mapping" is a >>> hardware property. What about: >>> >>> Subject: pinctrl: imx1-core: Fix debug output in .pin_config_set callback >>> >>> imx1_pinconf_set assumes that the array of pins in struct >>> imx1_pinctrl_soc_info can be indexed by pin id to get the >>> pinctrl_pin_desc for a pin. This used to be correct up to commit >>> 607af165c047 ("pinctrl: i.MX27: Remove nonexistent pad definitions") >>> which removed some entries from the array and so made it wrong to access >>> the array by pin id. >>> >>> Implement the easiest fix by not resolving the pin id to a name but >>> printing the id instead. >>> >>> Fixes: 607af165c047 ("pinctrl: i.MX27: Remove nonexistent pad definitions") >>> Cc: stable@vger.kernel.org >>> Reported-by: Chris Ruehl <chris.ruehl@gtsys.com.hk> >>> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> >>> >>> Having the pad name in the output is nice, is it worth to search for the >>> right pinctrl_pin_desc in the array? The array is still sorted, so a >>> binary search would do, maybe a function for this already exists? >>> >>> How did Chris notice the error? Just a bogus output, or did it crash the >>> kernel? That would be worth to note in the commit log, too. >>> >>> Otherwise the change looks fine. >>> >>> Best regards >>> Uwe >> >> I had a crash on a array overrun. >> >> I'd applied a patch to the linux-usb (which was spit out by the maintainer >> perl script) > > Yes, as there is already a patch from Chris my patch can be dropped. We > should continue the discussion on your patch instead. > >> chris@wheezyvm:~/kernel.d/linux-next$ cat >> 0001-Pinctrl-imx1-fix-wrong-pin-name-resolving.patch >> From 50d56e5f626b2ea86211818cf487514c96f60487 Mon Sep 17 00:00:00 2001 >> From: Chris Ruehl <chris.ruehl@gtsys.com.hk> >> Date: Sat, 23 May 2015 15:02:44 +0800 >> Subject: [PATCH] Pinctrl: imx1: fix wrong pin-name resolving >> >> Bug in function imx1_pinconf_set() cause crash when >> princtrl debug is enabled and the pin_id becomes larger >> then the info->pins[] contains. >> >> imx27-pinctrl 10015000.iomuxc: request pin 134 (MX27_PAD_UART2_TXD) for >> 1000b000.serial >> imx27-pinctrl 10015000.iomuxc: request pin 135 (MX27_PAD_UART2_RXD) for >> 1000b000.serial >> imx27-pinctrl 10015000.iomuxc: request pin 131 (MX27_PAD_UART2_CTS) for >> 1000b000.serial >> imx27-pinctrl 10015000.iomuxc: request pin 132 (MX27_PAD_UART2_RTS) for >> 1000b000.serial >> imx27-pinctrl 10015000.iomuxc: enable function uart group uart2-1 >> imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x86, function 0, gpio 0, >> direction 1, oconf 0, iconfa 0, iconfb 0 >> imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x87, function 0, gpio 0, >> direction 0, oconf 0, iconfa 0, iconfb 0 >> imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x83, function 0, gpio 0, >> direction 1, oconf 0, iconfa 0, iconfb 0 >> imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0x84, function 0, gpio 0, >> direction 0, oconf 0, iconfa 0, iconfb 0 >> imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=134 >> imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_UART1_RTS >> imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=135 >> imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_RTCK >> imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=131 >> imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_UART1_TXD >> imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=132 >> imx27-pinctrl 10015000.iomuxc: pinconf set pullup pin MX27_PAD_UART1_RXD >> ... >> imx27-pinctrl 10015000.iomuxc: write: register 0xf4415508 offset 4 value 0x3 >> imx27-pinctrl 10015000.iomuxc: write: register 0xf4415510 offset 4 value 0x0 >> imx27-pinctrl 10015000.iomuxc: write: register 0xf4415518 offset 4 value 0x0 >> imx27-pinctrl 10015000.iomuxc: imx1_pmx_set, pin 0xb5, function 0, gpio 0, >> direction 1, oconf 0, iconfa 0, iconfb 0 >> imx27-pinctrl 10015000.iomuxc: num_configs=1 PinID=171 >> Unable to handle kernel paging request at virtual address 6c61765f >> pgd = c0004000 >> 6c61765f] *pgd=00000000 >> Internal error: Oops: 5 [#1] ARM >> Modules linked in: >> CPU: 0 PID: 1 Comm: swapper Not tainted 4.1.0-rc4-next-20150522-dirty #8 >> Hardware name: GTSYS i.MX27GTSIR (Device Tree Support) >> task: ce832000 ti: ce848000 task.ti: ce848000 >> PC is at strnlen+0x28/0x3c >> LR is at string.isra.4+0x34/0xcc >> pc : [<c01ae188>] lr : [<c01af9a4>] psr: 20000093 >> sp : ce849a88 ip : ce849a98 fp : ce849a94 >> r10: c05d7e3a r9 : c05d81e4 r8 : 00000000 >> r7 : 6c61765f r6 : c05d81e4 r5 : ffffffff r4 : c05d7e3a >> r3 : 6c61765f r2 : 6c61765f r1 : 6c61765e r0 : 6c61765f >> Flags: nzCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel >> >> >> Signed-off-by: Chris Ruehl <chris.ruehl@gtsys.com.hk> >> --- >> drivers/pinctrl/freescale/pinctrl-imx1-core.c | 6 ++++-- >> 1 file changed, 4 insertions(+), 2 deletions(-) >> >> diff --git a/drivers/pinctrl/freescale/pinctrl-imx1-core.c >> b/drivers/pinctrl/freescale/pinctrl-imx1-core.c >> index 5ac59fb..8408bd8 100644 >> --- a/drivers/pinctrl/freescale/pinctrl-imx1-core.c >> +++ b/drivers/pinctrl/freescale/pinctrl-imx1-core.c >> @@ -403,14 +403,16 @@ static int imx1_pinconf_set(struct pinctrl_dev *pctldev, >> unsigned num_configs) >> { >> struct imx1_pinctrl *ipctl = pinctrl_dev_get_drvdata(pctldev); >> - const struct imx1_pinctrl_soc_info *info = ipctl->info; >> + struct pin_desc *desc; >> int i; >> >> + desc = pin_desc_get(pctldev, pin_id); > > For the rest of this function pin_desc_get() is not necessary. This is > not a simple function call but a radix tree lookup as far as I can see. > I am not sure how important this debugging information is and how > expensive this function call is?! > > Best Regards, > > Markus > Hi Markus, I found it helpful to debug my boards configuration, and have the pin debug on, then I see the output only while the kernel initiation phase. If you look around, the pin_desc_get() function is used quite often in other parts of the pinctrl section. drivers/pinctrl/pinctrl-falcon.c drivers/pinctrl/pinconf.c drivers/pinctrl/pinmux.c drivers/pinctrl/core.c (me) drivers/pinctrl/freescale/pinctrl-imx1-core.c drivers/pinctrl/pinctrl-single.c Regards Chris
Hi, On Thu, Jul 09, 2015 at 08:06:52PM +0200, Uwe Kleine-König wrote: > Hi, > > On Thu, Jul 09, 2015 at 03:30:36PM +0200, Markus Pargmann wrote: > > > b/drivers/pinctrl/freescale/pinctrl-imx1-core.c > > > index 5ac59fb..8408bd8 100644 > > > --- a/drivers/pinctrl/freescale/pinctrl-imx1-core.c > > > +++ b/drivers/pinctrl/freescale/pinctrl-imx1-core.c > > > @@ -403,14 +403,16 @@ static int imx1_pinconf_set(struct pinctrl_dev *pctldev, > > > unsigned num_configs) > > > { > > > struct imx1_pinctrl *ipctl = pinctrl_dev_get_drvdata(pctldev); > > > - const struct imx1_pinctrl_soc_info *info = ipctl->info; > > > + struct pin_desc *desc; > > > int i; > > > > > > + desc = pin_desc_get(pctldev, pin_id); > > > > For the rest of this function pin_desc_get() is not necessary. This is > > not a simple function call but a radix tree lookup as far as I can see. > > I am not sure how important this debugging information is and how > > expensive this function call is?! > If it's moved into the call to pr_debug it should be optimized away when > debugging is off. That together with a merge of Chris' and mine commit > log and I'm happy. Ok. If you change that I am happy as well. Best Regards, Markus
diff --git a/drivers/pinctrl/freescale/pinctrl-imx1-core.c b/drivers/pinctrl/freescale/pinctrl-imx1-core.c index 5ac59fb..8408bd8 100644 --- a/drivers/pinctrl/freescale/pinctrl-imx1-core.c +++ b/drivers/pinctrl/freescale/pinctrl-imx1-core.c @@ -403,14 +403,16 @@ static int imx1_pinconf_set(struct pinctrl_dev *pctldev, unsigned num_configs) { struct imx1_pinctrl *ipctl = pinctrl_dev_get_drvdata(pctldev); - const struct imx1_pinctrl_soc_info *info = ipctl->info; + struct pin_desc *desc; int i; + desc = pin_desc_get(pctldev, pin_id); + for (i = 0; i != num_configs; ++i) { imx1_write_bit(ipctl, pin_id, configs[i] & 0x01, MX1_PUEN); dev_dbg(ipctl->dev, "pinconf set pullup pin %s\n", - info->pins[pin_id].name); + desc->name); } return 0;