ChangeSet 1.2223.2.5, 2004/11/24 14:25:49-08:00, aris@cathedrallabs.org [PATCH] i2c-elektor: get rid of cli/sti this patch get rid of cli()/sti(). while correcting this I found that when a process wakes by an interrupt, pcf_pending doesn't come back to 0 and next caller will return imediately. also, there are other drivers with the exact same problem. if you don't have any comments on this, I'll do the same for other drivers. [I2C] i2c-elektor: getting rid of cli()/sti() usage Signed-off-by: Aristeu Sergio Rozanski Filho Signed-off-by: Greg Kroah-Hartman drivers/i2c/busses/i2c-elektor.c | 22 ++++++++++++++++++---- 1 files changed, 18 insertions(+), 4 deletions(-) diff -Nru a/drivers/i2c/busses/i2c-elektor.c b/drivers/i2c/busses/i2c-elektor.c --- a/drivers/i2c/busses/i2c-elektor.c 2004-11-30 16:01:05 -08:00 +++ b/drivers/i2c/busses/i2c-elektor.c 2004-11-30 16:01:05 -08:00 @@ -59,6 +59,7 @@ static wait_queue_head_t pcf_wait; static int pcf_pending; +static spinlock_t lock; /* ----- local functions ---------------------------------------------- */ @@ -111,14 +112,24 @@ static void pcf_isa_waitforpin(void) { int timeout = 2; + long flags; if (irq > 0) { - cli(); + spin_lock_irqsave(&lock, flags); if (pcf_pending == 0) { - interruptible_sleep_on_timeout(&pcf_wait, timeout*HZ ); - } else + spin_unlock_irqrestore(&lock, flags); + if (interruptible_sleep_on_timeout(&pcf_wait, + timeout*HZ)) { + spin_lock_irqsave(&lock, flags); + if (pcf_pending == 1) { + pcf_pending = 0; + } + spin_unlock_irqrestore(&lock, flags); + } + } else { pcf_pending = 0; - sti(); + spin_unlock_irqrestore(&lock, flags); + } } else { udelay(100); } @@ -126,7 +137,9 @@ static irqreturn_t pcf_isa_handler(int this_irq, void *dev_id, struct pt_regs *regs) { + spin_lock(&lock); pcf_pending = 1; + spin_unlock(&lock); wake_up_interruptible(&pcf_wait); return IRQ_HANDLED; } @@ -134,6 +147,7 @@ static int pcf_isa_init(void) { + spin_lock_init(&lock); if (!mmapped) { if (!request_region(base, 2, "i2c (isa bus adapter)")) { printk(KERN_ERR