* [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver
@ 2008-03-27 14:00 Sebastian Siewior
2008-03-27 14:00 ` [PATCH / RFC 1/2] fec: kill warnings Sebastian Siewior
` (4 more replies)
0 siblings, 5 replies; 12+ messages in thread
From: Sebastian Siewior @ 2008-03-27 14:00 UTC (permalink / raw)
To: uclinux-dev; +Cc: netdev, Jeff Garzik, Greg Ungerer
I got into this after turning on spinlock debugging.
Jeff: The dev->hard_start_xmit callback is always called from softirq
context so it is save to use spin_lock_irq(), right?
Greg: The isr (fec_enet_rx()) calls three functions. All
but fec_enet_rx() are grabbing the spinlock. Is this on purpose?
Sebastian
--
^ permalink raw reply [flat|nested] 12+ messages in thread* [PATCH / RFC 1/2] fec: kill warnings 2008-03-27 14:00 [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Sebastian Siewior @ 2008-03-27 14:00 ` Sebastian Siewior 2008-03-27 14:00 ` [PATCH / RFC 2/2] fec: fixup spinlocks Sebastian Siewior ` (3 subsequent siblings) 4 siblings, 0 replies; 12+ messages in thread From: Sebastian Siewior @ 2008-03-27 14:00 UTC (permalink / raw) To: uclinux-dev; +Cc: netdev, Jeff Garzik, Greg Ungerer [-- Attachment #1: m68knommu-remove_unused_in_fec.patch --] [-- Type: text/plain, Size: 1566 bytes --] linux-2.6/drivers/net/fec.c: In function 'fec_enet_module_init': linux-2.6/drivers/net/fec.c:2627: warning: unused variable 'j' linux-2.6/drivers/net/fec.c: At top level: linux-2.6/drivers/net/fec.c:2136: warning: 'mii_link_interrupt' defined but not used Signed-off-by: Sebastian Siewior <bigeasy@linutronix.de> --- a/drivers/net/fec.c +++ b/drivers/net/fec.c @@ -67,6 +67,11 @@ #define FEC_MAX_PORTS 1 #endif +#if defined(CONFIG_M523x) || defined(CONFIG_M527x) || defined(CONFIG_M528x) || \ + defined(CONFIG_M520x) || defined(CONFIG_M532x) +#define DONT_NEED_mii_link_interrupt +#endif + /* * Define the fixed address of the FEC hardware. */ @@ -1222,7 +1227,7 @@ static phy_info_t const * const phy_info }; /* ------------------------------------------------------------------------- */ -#if !defined(CONFIG_M532x) +#ifndef DONT_NEED_mii_link_interrupt #ifdef CONFIG_RPXCLASSIC static void mii_link_interrupt(void *dev_id); @@ -2126,6 +2131,7 @@ mii_discover_phy(uint mii_reg, struct ne /* This interrupt occurs when the PHY detects a link change. */ +#ifndef DONT_NEED_mii_link_interrupt #ifdef CONFIG_RPXCLASSIC static void mii_link_interrupt(void *dev_id) @@ -2148,6 +2154,7 @@ mii_link_interrupt(int irq, void * dev_i return IRQ_HANDLED; } +#endif static int fec_enet_open(struct net_device *dev) @@ -2624,7 +2631,7 @@ fec_stop(struct net_device *dev) static int __init fec_enet_module_init(void) { struct net_device *dev; - int i, j, err; + int i, err; DECLARE_MAC_BUF(mac); printk("FEC ENET Version 0.2\n"); -- ^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH / RFC 2/2] fec: fixup spinlocks. 2008-03-27 14:00 [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Sebastian Siewior 2008-03-27 14:00 ` [PATCH / RFC 1/2] fec: kill warnings Sebastian Siewior @ 2008-03-27 14:00 ` Sebastian Siewior 2008-03-28 6:58 ` Greg Ungerer 2008-03-28 6:51 ` [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Greg Ungerer ` (2 subsequent siblings) 4 siblings, 1 reply; 12+ messages in thread From: Sebastian Siewior @ 2008-03-27 14:00 UTC (permalink / raw) To: uclinux-dev; +Cc: netdev, Jeff Garzik, Greg Ungerer [-- Attachment #1: m68knommu-fix_spinlocks_in_fec.patch --] [-- Type: text/plain, Size: 4197 bytes --] 1. Initialize the spinlock 2. Fix the following spinlock recursion: |BUG: spinlock recursion on CPU#0, swapper/1 | lock: 00253484, .magic: dead4ead, .owner: swapper/1, .owner_cpu: 0 |Stack from 00219d04: | 0002951e 00144b04 000afaca 00253484 001753a2 00002704 000c5ef6 00253484 | 00800000 0025302d 002533e0 0002951e 00253000 0004857a 00144baa 00253484 | 00000000 000c5550 00253484 00172acd 000c5ef6 0017689c 0014aca3 00253484 | 00020013 00252000 002533e0 40001000 000c602e 00253000 600e0000 000c5ef6 | 001a6d68 40001000 000c57c0 00020013 00253000 00172acd 000c5faa 00176a11 | 0014ac96 000c5faa 00000000 00000000 0000005d 00000033 00219eaa 00000001 |Possible Call Trace: | printk+0x0/0x20 | _spin_unlock+0x0/0x6 | _raw_spin_lock+0xfc/0x160 | mii_discover_phy3+0x0/0xb4 | printk+0x0/0x20 | note_interrupt+0x0/0x28a | _spin_lock_irqsave+0x16/0x1e | mii_queue+0x4c/0x124 | mii_discover_phy3+0x0/0xb4 | _start+0x13/0x7c | mii_discover_phy+0x84/0x90 | mii_discover_phy3+0x0/0xb4 | fec_enet_interrupt+0x158/0x472 We call from inet context (fec_enet_interrupt, holding the spinlock) the first callback (which is mii_discover_phy(), initialized in fec_enet_init()). mii_discover_phy() calls mii_queue() which takes the spinlock again, boom. The fix is to drop the spinlock in interrupt context after the list modification. 3. Use spin_unlock_irq / spin_lock_irq in the IRQ functions (IRQ is not registered with the IRQ off flag) and dev->hard_start_xmit callback. 4. Use the .*_irqsave variant in that part which may be called from IRQ or user mode. Signed-off-by: Sebastian Siewior <bigeasy@linutronix.de> --- a/drivers/net/fec.c +++ b/drivers/net/fec.c @@ -117,9 +117,11 @@ static unsigned char fec_mac_default[] = /* Forward declarations of some structures to support different PHYs */ +typedef void (mii_func)(uint val, struct net_device *dev); + typedef struct { uint mii_data; - void (*funct)(uint mii_reg, struct net_device *dev); + mii_func *funct; } phy_cmd_t; typedef struct { @@ -261,7 +263,7 @@ static mii_list_t *mii_head; static mii_list_t *mii_tail; static int mii_queue(struct net_device *dev, int request, - void (*func)(uint, struct net_device *)); + mii_func *func); /* Make MII read/write commands for the FEC. */ @@ -503,7 +505,7 @@ fec_enet_tx(struct net_device *dev) struct sk_buff *skb; fep = netdev_priv(dev); - spin_lock(&fep->lock); + spin_lock_irq(&fep->lock); bdp = fep->dirty_tx; while (((status = bdp->cbd_sc) & BD_ENET_TX_READY) == 0) { @@ -562,7 +564,7 @@ fec_enet_tx(struct net_device *dev) } } fep->dirty_tx = (cbd_t *)bdp; - spin_unlock(&fep->lock); + spin_unlock_irq(&fep->lock); } @@ -705,12 +707,13 @@ fec_enet_mii(struct net_device *dev) volatile fec_t *ep; mii_list_t *mip; uint mii_reg; + mii_func *mii_func = NULL; fep = netdev_priv(dev); ep = fep->hwp; mii_reg = ep->fec_mii_data; - spin_lock(&fep->lock); + spin_lock_irq(&fep->lock); if ((mip = mii_head) == NULL) { printk("MII and no head!\n"); @@ -718,7 +721,7 @@ fec_enet_mii(struct net_device *dev) } if (mip->mii_func != NULL) - (*(mip->mii_func))(mii_reg, dev); + mii_func = *(mip->mii_func); mii_head = mip->mii_next; mip->mii_next = mii_free; @@ -728,11 +731,13 @@ fec_enet_mii(struct net_device *dev) ep->fec_mii_data = mip->mii_regval; unlock: - spin_unlock(&fep->lock); + spin_unlock_irq(&fep->lock); + if (mii_func) + mii_func(mii_reg, dev); } static int -mii_queue(struct net_device *dev, int regval, void (*func)(uint, struct net_device *)) +mii_queue(struct net_device *dev, int regval, mii_func *func) { struct fec_enet_private *fep; unsigned long flags; @@ -2343,6 +2348,7 @@ int __init fec_enet_init(struct net_devi */ fecp = (volatile fec_t *) fec_hw[index]; + spin_lock_init(&fep->lock); fep->index = index; fep->hwp = fecp; fep->netdev = dev; -- _______________________________________________ uClinux-dev mailing list uClinux-dev@uclinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev@uclinux.org To unsubscribe see: http://mailman.uclinux.org/mailman/options/uclinux-dev ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 2/2] fec: fixup spinlocks. 2008-03-27 14:00 ` [PATCH / RFC 2/2] fec: fixup spinlocks Sebastian Siewior @ 2008-03-28 6:58 ` Greg Ungerer 2008-03-28 9:24 ` Sebastian Siewior 0 siblings, 1 reply; 12+ messages in thread From: Greg Ungerer @ 2008-03-28 6:58 UTC (permalink / raw) To: Sebastian Siewior; +Cc: uclinux-dev, netdev, Jeff Garzik Hi Sebastion, Sebastian Siewior wrote: > 1. Initialize the spinlock > 2. Fix the following spinlock recursion: > > |BUG: spinlock recursion on CPU#0, swapper/1 > | lock: 00253484, .magic: dead4ead, .owner: swapper/1, .owner_cpu: 0 > |Stack from 00219d04: > | 0002951e 00144b04 000afaca 00253484 001753a2 00002704 000c5ef6 00253484 > | 00800000 0025302d 002533e0 0002951e 00253000 0004857a 00144baa 00253484 > | 00000000 000c5550 00253484 00172acd 000c5ef6 0017689c 0014aca3 00253484 > | 00020013 00252000 002533e0 40001000 000c602e 00253000 600e0000 000c5ef6 > | 001a6d68 40001000 000c57c0 00020013 00253000 00172acd 000c5faa 00176a11 > | 0014ac96 000c5faa 00000000 00000000 0000005d 00000033 00219eaa 00000001 > |Possible Call Trace: > | printk+0x0/0x20 > | _spin_unlock+0x0/0x6 > | _raw_spin_lock+0xfc/0x160 > | mii_discover_phy3+0x0/0xb4 > | printk+0x0/0x20 > | note_interrupt+0x0/0x28a > | _spin_lock_irqsave+0x16/0x1e > | mii_queue+0x4c/0x124 > | mii_discover_phy3+0x0/0xb4 > | _start+0x13/0x7c > | mii_discover_phy+0x84/0x90 > | mii_discover_phy3+0x0/0xb4 > | fec_enet_interrupt+0x158/0x472 > > We call from inet context (fec_enet_interrupt, holding the spinlock) the > first callback (which is mii_discover_phy(), initialized in fec_enet_init()). > mii_discover_phy() calls mii_queue() which takes the spinlock again, boom. > The fix is to drop the spinlock in interrupt context after the list > modification. > > 3. Use spin_unlock_irq / spin_lock_irq in the IRQ functions (IRQ is not > registered with the IRQ off flag) and dev->hard_start_xmit callback. > 4. Use the .*_irqsave variant in that part which may be called from IRQ or > user mode. I couldn't see any changes here that switched to use *_irqsave? Regards Greg > Signed-off-by: Sebastian Siewior <bigeasy@linutronix.de> > --- a/drivers/net/fec.c > +++ b/drivers/net/fec.c > @@ -117,9 +117,11 @@ static unsigned char fec_mac_default[] = > /* Forward declarations of some structures to support different PHYs > */ > > +typedef void (mii_func)(uint val, struct net_device *dev); > + > typedef struct { > uint mii_data; > - void (*funct)(uint mii_reg, struct net_device *dev); > + mii_func *funct; > } phy_cmd_t; > > typedef struct { > @@ -261,7 +263,7 @@ static mii_list_t *mii_head; > static mii_list_t *mii_tail; > > static int mii_queue(struct net_device *dev, int request, > - void (*func)(uint, struct net_device *)); > + mii_func *func); > > /* Make MII read/write commands for the FEC. > */ > @@ -503,7 +505,7 @@ fec_enet_tx(struct net_device *dev) > struct sk_buff *skb; > > fep = netdev_priv(dev); > - spin_lock(&fep->lock); > + spin_lock_irq(&fep->lock); > bdp = fep->dirty_tx; > > while (((status = bdp->cbd_sc) & BD_ENET_TX_READY) == 0) { > @@ -562,7 +564,7 @@ fec_enet_tx(struct net_device *dev) > } > } > fep->dirty_tx = (cbd_t *)bdp; > - spin_unlock(&fep->lock); > + spin_unlock_irq(&fep->lock); > } > > > @@ -705,12 +707,13 @@ fec_enet_mii(struct net_device *dev) > volatile fec_t *ep; > mii_list_t *mip; > uint mii_reg; > + mii_func *mii_func = NULL; > > fep = netdev_priv(dev); > ep = fep->hwp; > mii_reg = ep->fec_mii_data; > > - spin_lock(&fep->lock); > + spin_lock_irq(&fep->lock); > > if ((mip = mii_head) == NULL) { > printk("MII and no head!\n"); > @@ -718,7 +721,7 @@ fec_enet_mii(struct net_device *dev) > } > > if (mip->mii_func != NULL) > - (*(mip->mii_func))(mii_reg, dev); > + mii_func = *(mip->mii_func); > > mii_head = mip->mii_next; > mip->mii_next = mii_free; > @@ -728,11 +731,13 @@ fec_enet_mii(struct net_device *dev) > ep->fec_mii_data = mip->mii_regval; > > unlock: > - spin_unlock(&fep->lock); > + spin_unlock_irq(&fep->lock); > + if (mii_func) > + mii_func(mii_reg, dev); > } > > static int > -mii_queue(struct net_device *dev, int regval, void (*func)(uint, struct net_device *)) > +mii_queue(struct net_device *dev, int regval, mii_func *func) > { > struct fec_enet_private *fep; > unsigned long flags; > @@ -2343,6 +2348,7 @@ int __init fec_enet_init(struct net_devi > */ > fecp = (volatile fec_t *) fec_hw[index]; > > + spin_lock_init(&fep->lock); > fep->index = index; > fep->hwp = fecp; > fep->netdev = dev; > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg@snapgear.com Secure Computing Corporation PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 2/2] fec: fixup spinlocks. 2008-03-28 6:58 ` Greg Ungerer @ 2008-03-28 9:24 ` Sebastian Siewior 2008-03-28 10:24 ` Greg Ungerer 0 siblings, 1 reply; 12+ messages in thread From: Sebastian Siewior @ 2008-03-28 9:24 UTC (permalink / raw) To: Greg Ungerer; +Cc: uclinux-dev, netdev, Jeff Garzik * Greg Ungerer | 2008-03-28 16:58:09 [+1000]: >Hi Sebastion, Hi Greg, >>4. Use the .*_irqsave variant in that part which may be called from IRQ or >>user mode. > >I couldn't see any changes here that switched to use *_irqsave? The only lock user that may be called from both called from both places is mii_queue() and that one allready uses the save variant. I guess you want me to repost that patch without point 4 in the summary right? >Regards >Greg Sebastian ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 2/2] fec: fixup spinlocks. 2008-03-28 9:24 ` Sebastian Siewior @ 2008-03-28 10:24 ` Greg Ungerer 0 siblings, 0 replies; 12+ messages in thread From: Greg Ungerer @ 2008-03-28 10:24 UTC (permalink / raw) To: Sebastian Siewior; +Cc: netdev, Jeff Garzik, uclinux-dev Hi Sebastion, Sebastian Siewior wrote: > * Greg Ungerer | 2008-03-28 16:58:09 [+1000]: > >> Hi Sebastion, > Hi Greg, > >>> 4. Use the .*_irqsave variant in that part which may be called from IRQ or >>> user mode. >> I couldn't see any changes here that switched to use *_irqsave? > The only lock user that may be called from both called from both places > is mii_queue() and that one allready uses the save variant. > I guess you want me to repost that patch without point 4 in the summary > right? No, don't worry about it. I just wanted to make sure there wasn't more to the patch. Thanks Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg@snapgear.com SnapGear -- a Secure Computing Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com _______________________________________________ uClinux-dev mailing list uClinux-dev@uclinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev@uclinux.org To unsubscribe see: http://mailman.uclinux.org/mailman/options/uclinux-dev ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver 2008-03-27 14:00 [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Sebastian Siewior 2008-03-27 14:00 ` [PATCH / RFC 1/2] fec: kill warnings Sebastian Siewior 2008-03-27 14:00 ` [PATCH / RFC 2/2] fec: fixup spinlocks Sebastian Siewior @ 2008-03-28 6:51 ` Greg Ungerer 2008-04-02 11:28 ` Sebastian Siewior 2008-03-29 2:19 ` Jeff Garzik 2008-03-29 2:21 ` Jeff Garzik 4 siblings, 1 reply; 12+ messages in thread From: Greg Ungerer @ 2008-03-28 6:51 UTC (permalink / raw) To: Sebastian Siewior; +Cc: uclinux-dev, netdev, Jeff Garzik Hi Sebastion, Sebastian Siewior wrote: > I got into this after turning on spinlock debugging. > > Jeff: The dev->hard_start_xmit callback is always called from softirq > context so it is save to use spin_lock_irq(), right? > > Greg: The isr (fec_enet_rx()) calls three functions. All (fec_enet_interrupt()) > but fec_enet_rx() are grabbing the spinlock. Is this on purpose? No, it looks broken :-( Regards Greg -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg@snapgear.com Secure Computing Corporation PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver 2008-03-28 6:51 ` [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Greg Ungerer @ 2008-04-02 11:28 ` Sebastian Siewior 2008-04-02 12:50 ` Greg Ungerer 0 siblings, 1 reply; 12+ messages in thread From: Sebastian Siewior @ 2008-04-02 11:28 UTC (permalink / raw) To: Greg Ungerer; +Cc: uclinux-dev, netdev, Jeff Garzik Greg Ungerer wrote: > Hi Sebastion, s/o/a :) > No, it looks broken :-( I'm going to post a cleanup patch and a locking fixup patch. The missing lock is more serious than I though in first place. > Regards > Greg Sebastian ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver 2008-04-02 11:28 ` Sebastian Siewior @ 2008-04-02 12:50 ` Greg Ungerer 0 siblings, 0 replies; 12+ messages in thread From: Greg Ungerer @ 2008-04-02 12:50 UTC (permalink / raw) To: Sebastian Siewior; +Cc: uclinux-dev, netdev, Jeff Garzik Sebastian Siewior wrote: > Greg Ungerer wrote: >> Hi Sebastion, > s/o/a :) Sorry :-) >> No, it looks broken :-( > I'm going to post a cleanup patch and a locking fixup patch. The missing > lock is more serious than I though in first place. Be aware there are other known problems with this driver "as is". Search for the recent discussion on uclinux-dev mailing list about it. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg@snapgear.com SnapGear -- a Secure Computing Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver 2008-03-27 14:00 [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Sebastian Siewior ` (2 preceding siblings ...) 2008-03-28 6:51 ` [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Greg Ungerer @ 2008-03-29 2:19 ` Jeff Garzik 2008-03-29 2:21 ` Jeff Garzik 4 siblings, 0 replies; 12+ messages in thread From: Jeff Garzik @ 2008-03-29 2:19 UTC (permalink / raw) To: Sebastian Siewior; +Cc: uclinux-dev, netdev, Greg Ungerer Sebastian Siewior wrote: > I got into this after turning on spinlock debugging. > > Jeff: The dev->hard_start_xmit callback is always called from softirq > context so it is save to use spin_lock_irq(), right? Some drivers do that, and it's fine. But be aware spin_lock_irqsave() may be required for edge cases like netconsole. ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver 2008-03-27 14:00 [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Sebastian Siewior ` (3 preceding siblings ...) 2008-03-29 2:19 ` Jeff Garzik @ 2008-03-29 2:21 ` Jeff Garzik 2008-04-02 11:11 ` Sebastian Siewior 4 siblings, 1 reply; 12+ messages in thread From: Jeff Garzik @ 2008-03-29 2:21 UTC (permalink / raw) To: Sebastian Siewior; +Cc: uclinux-dev, netdev, Greg Ungerer Sebastian Siewior wrote: > I got into this after turning on spinlock debugging. > > Jeff: The dev->hard_start_xmit callback is always called from softirq > context so it is save to use spin_lock_irq(), right? > > Greg: The isr (fec_enet_rx()) calls three functions. All > but fec_enet_rx() are grabbing the spinlock. Is this on purpose? wanted for 2.6.25? ACKs from maintainer? ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver 2008-03-29 2:21 ` Jeff Garzik @ 2008-04-02 11:11 ` Sebastian Siewior 0 siblings, 0 replies; 12+ messages in thread From: Sebastian Siewior @ 2008-04-02 11:11 UTC (permalink / raw) To: Jeff Garzik; +Cc: Sebastian Siewior, uclinux-dev, netdev, Greg Ungerer * Jeff Garzik | 2008-03-28 22:21:05 [-0400]: > Sebastian Siewior wrote: >> I got into this after turning on spinlock debugging. >> Jeff: The dev->hard_start_xmit callback is always called from softirq >> context so it is save to use spin_lock_irq(), right? >> Greg: The isr (fec_enet_rx()) calls three functions. All >> but fec_enet_rx() are grabbing the spinlock. Is this on purpose? Jeff please drop that one, I form a new patch just located another BUG. That one triggers on high load on a non-smp, non-preempt machine. Sebastian ^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2008-04-02 12:50 UTC | newest] Thread overview: 12+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2008-03-27 14:00 [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Sebastian Siewior 2008-03-27 14:00 ` [PATCH / RFC 1/2] fec: kill warnings Sebastian Siewior 2008-03-27 14:00 ` [PATCH / RFC 2/2] fec: fixup spinlocks Sebastian Siewior 2008-03-28 6:58 ` Greg Ungerer 2008-03-28 9:24 ` Sebastian Siewior 2008-03-28 10:24 ` Greg Ungerer 2008-03-28 6:51 ` [PATCH / RFC 0/2] spinlocks fixup in fec / m68knommu driver Greg Ungerer 2008-04-02 11:28 ` Sebastian Siewior 2008-04-02 12:50 ` Greg Ungerer 2008-03-29 2:19 ` Jeff Garzik 2008-03-29 2:21 ` Jeff Garzik 2008-04-02 11:11 ` Sebastian Siewior
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).