[PATCH 0/2]linux-next: fix two bugs in ACPI EC driver

[PATCH 0/2]linux-next: fix two bugs in ACPI EC driver

[Kevin Hao]

Hi all,

These patches fix two bugs in ACPI EC driver.
First one fix ec read and write data bug.
Second one fix ec EC_FLAGS_GPE_STORM flag always set bug.

Thanks,
Kevin

[PATCH 1/2] fix acpi ec read write bug

[Kevin Hao]

Fill in command unit of transaction_data structure, otherwise
gpe_transaction will skip read or write instruction.

Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
---
 drivers/acpi/ec.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
index bd3555c..0c65e82 100644
--- a/drivers/acpi/ec.c
+++ b/drivers/acpi/ec.c
@@ -240,6 +240,7 @@ static int acpi_ec_transaction_unlocked(struct acpi_ec *ec, u8 command,
 	}
 	atomic_set(&ec->irq_count, 0);
 	/* fill in transaction structure */
+	ec->t.command = command;
 	ec->t.wdata = wdata;
 	ec->t.wlen = wdata_len;
 	ec->t.rdata = rdata;
-- 
1.5.6.2.220.g44701

[PATCH 2/2] fix acpi ec set GPE storm flag bug

[Kevin Hao]

The intent of the code was clear, however the missing braces
meant that EC_FLAGS_GPE_STORM was incorrectly being set
all the time

Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
---
 drivers/acpi/ec.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
index 0c65e82..d8f273d 100644
--- a/drivers/acpi/ec.c
+++ b/drivers/acpi/ec.c
@@ -268,9 +268,10 @@ end:
 		/* it is safe to enable GPE outside of transaction */
 		acpi_enable_gpe(NULL, ec->gpe, ACPI_NOT_ISR);
 	} else if (test_bit(EC_FLAGS_GPE_MODE, &ec->flags) &&
-		   atomic_read(&ec->irq_count) > ACPI_EC_STORM_THRESHOLD)
+		   atomic_read(&ec->irq_count) > ACPI_EC_STORM_THRESHOLD) {
 		pr_debug(PREFIX "GPE storm detected\n");
 		set_bit(EC_FLAGS_GPE_STORM, &ec->flags);
+	}
 	return 0;
 }
 
-- 
1.5.6.2.220.g44701

Re: [PATCH 1/2] fix acpi ec read write bug

[Zhao Yakui]

On Thu, 2008-09-11 at 11:17 +0800, Kevin Hao wrote:
> Fill in command unit of transaction_data structure, otherwise
> gpe_transaction will skip read or write instruction.
> 
> Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
> ---
>  drivers/acpi/ec.c |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
> index bd3555c..0c65e82 100644
> --- a/drivers/acpi/ec.c
> +++ b/drivers/acpi/ec.c
> @@ -240,6 +240,7 @@ static int acpi_ec_transaction_unlocked(struct acpi_ec *ec, u8 command,
>  	}
>  	atomic_set(&ec->irq_count, 0);
>  	/* fill in transaction structure */
> +	ec->t.command = command;
It is also OK to add this explicitly. In fact the ec->t.command will be
assigned in the function of acpi_ec_write_cmd.

Thanks.
>  	ec->t.wdata = wdata;
>  	ec->t.wlen = wdata_len;
>  	ec->t.rdata = rdata;

Re: [PATCH 1/2] fix acpi ec read write bug

[Kevin Hao]

On Thu, 2008-09-11 at 12:07 +0800, Zhao Yakui wrote:
> On Thu, 2008-09-11 at 11:17 +0800, Kevin Hao wrote:
> > Fill in command unit of transaction_data structure, otherwise
> > gpe_transaction will skip read or write instruction.
> > 
> > Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
> > ---
> >  drivers/acpi/ec.c |    1 +
> >  1 files changed, 1 insertions(+), 0 deletions(-)
> > 
> > diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
> > index bd3555c..0c65e82 100644
> > --- a/drivers/acpi/ec.c
> > +++ b/drivers/acpi/ec.c
> > @@ -240,6 +240,7 @@ static int acpi_ec_transaction_unlocked(struct acpi_ec *ec, u8 command,
> >  	}
> >  	atomic_set(&ec->irq_count, 0);
> >  	/* fill in transaction structure */
> > +	ec->t.command = command;
> It is also OK to add this explicitly. In fact the ec->t.command will be
> assigned in the function of acpi_ec_write_cmd.

NO, I am using the latest linux-next kernel and I don't found
ec->t.command is assigned in acpi_ec_write_cmd function.

Thanks,
Kevin

> 
> Thanks.
> >  	ec->t.wdata = wdata;
> >  	ec->t.wlen = wdata_len;
> >  	ec->t.rdata = rdata;
> 

Re: [PATCH 1/2] fix acpi ec read write bug

[Zhao Yakui]

On Thu, 2008-09-11 at 12:49 +0800, Kevin Hao wrote:
> On Thu, 2008-09-11 at 12:07 +0800, Zhao Yakui wrote:
> > On Thu, 2008-09-11 at 11:17 +0800, Kevin Hao wrote:
> > > Fill in command unit of transaction_data structure, otherwise
> > > gpe_transaction will skip read or write instruction.
> > > 
> > > Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
> > > ---
> > >  drivers/acpi/ec.c |    1 +
> > >  1 files changed, 1 insertions(+), 0 deletions(-)
> > > 
> > > diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
> > > index bd3555c..0c65e82 100644
> > > --- a/drivers/acpi/ec.c
> > > +++ b/drivers/acpi/ec.c
> > > @@ -240,6 +240,7 @@ static int acpi_ec_transaction_unlocked(struct acpi_ec *ec, u8 command,
> > >  	}
> > >  	atomic_set(&ec->irq_count, 0);
> > >  	/* fill in transaction structure */
> > > +	ec->t.command = command;
> > It is also OK to add this explicitly. In fact the ec->t.command will be
> > assigned in the function of acpi_ec_write_cmd.
> 
> NO, I am using the latest linux-next kernel and I don't found
> ec->t.command is assigned in acpi_ec_write_cmd function.

static inline void acpi_ec_write_cmd(struct acpi_ec *ec, u8 command)
 {
 	pr_debug(PREFIX "<--- command = 0x%2.2x\n", command);
-	outb(command, ec->command_addr);
+	outb((ec->t.command = command), ec->command_addr);
 }

It is very obscure so that it is not easy to understand.
> Thanks,
> Kevin
> 
> > 
> > Thanks.
> > >  	ec->t.wdata = wdata;
> > >  	ec->t.wlen = wdata_len;
> > >  	ec->t.rdata = rdata;
> > 

Re: [PATCH 1/2] fix acpi ec read write bug

[Alexey Starikovskiy]

Kevin Hao wrote:
> Fill in command unit of transaction_data structure, otherwise
> gpe_transaction will skip read or write instruction.
> 
> Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
> ---
>  drivers/acpi/ec.c |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
> index bd3555c..0c65e82 100644
> --- a/drivers/acpi/ec.c
> +++ b/drivers/acpi/ec.c
> @@ -240,6 +240,7 @@ static int acpi_ec_transaction_unlocked(struct acpi_ec *ec, u8 command,
>  	}
>  	atomic_set(&ec->irq_count, 0);
>  	/* fill in transaction structure */
> +	ec->t.command = command;
>  	ec->t.wdata = wdata;
>  	ec->t.wlen = wdata_len;
>  	ec->t.rdata = rdata;
NAK
Command is filled in acpi_ec_write_command() in order to make window between start of
the transaction and validity of the transaction data to the bare minimum.