2008-02-15 20:59:33

by Jiri Slaby

[permalink] [raw]
Subject: [PATCH 1/3] WDEV: ath5k, fix lock imbalance

Omitted lock causes sparse warning
drivers/net/wireless/ath5k/base.c:1682:1: warning: context imbalance in 'ath5k_tasklet_rx' - different lock contexts for basic block

Add the lock to the guilty fail path.

Signed-off-by: Jiri Slaby <[email protected]>
Cc: Nick Kossifidis <[email protected]>
Cc: Luis R. Rodriguez <[email protected]>
---
drivers/net/wireless/ath5k/base.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/net/wireless/ath5k/base.c b/drivers/net/wireless/ath5k/base.c
index ddc8714..49d38e8 100644
--- a/drivers/net/wireless/ath5k/base.c
+++ b/drivers/net/wireless/ath5k/base.c
@@ -1715,6 +1715,7 @@ ath5k_tasklet_rx(unsigned long data)
break;
else if (unlikely(ret)) {
ATH5K_ERR(sc, "error in processing rx descriptor\n");
+ spin_unlock(&sc->rxbuflock);
return;
}

--
1.5.3.8


2008-02-15 20:59:59

by Jiri Slaby

[permalink] [raw]
Subject: [PATCH 2/3] WDEV, ath5k, don't return int from bool function

sparse sees int -> bool cast as an error:
hw.c:3754:10: warning: cast truncates bits from constant value (ffffffea becomes 0)
Fix it by converting the rettype to int and check appropriately.

Signed-off-by: Jiri Slaby <[email protected]>
Cc: Nick Kossifidis <[email protected]>
Cc: Luis R. Rodriguez <[email protected]>
---
drivers/net/wireless/ath5k/ath5k.h | 2 +-
drivers/net/wireless/ath5k/base.c | 5 ++++-
drivers/net/wireless/ath5k/hw.c | 8 ++++----
3 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/net/wireless/ath5k/ath5k.h b/drivers/net/wireless/ath5k/ath5k.h
index c79066b..69dea33 100644
--- a/drivers/net/wireless/ath5k/ath5k.h
+++ b/drivers/net/wireless/ath5k/ath5k.h
@@ -1035,7 +1035,7 @@ struct ath5k_hw {
unsigned int, unsigned int, enum ath5k_pkt_type, unsigned int,
unsigned int, unsigned int, unsigned int, unsigned int,
unsigned int, unsigned int, unsigned int);
- bool (*ah_setup_xtx_desc)(struct ath5k_hw *, struct ath5k_desc *,
+ int (*ah_setup_xtx_desc)(struct ath5k_hw *, struct ath5k_desc *,
unsigned int, unsigned int, unsigned int, unsigned int,
unsigned int, unsigned int);
int (*ah_proc_tx_desc)(struct ath5k_hw *, struct ath5k_desc *);
diff --git a/drivers/net/wireless/ath5k/base.c b/drivers/net/wireless/ath5k/base.c
index 49d38e8..59e5d56 100644
--- a/drivers/net/wireless/ath5k/base.c
+++ b/drivers/net/wireless/ath5k/base.c
@@ -668,7 +668,10 @@ ath5k_attach(struct pci_dev *pdev, struct ieee80211_hw *hw)
* return false w/o doing anything. MAC's that do
* support it will return true w/o doing anything.
*/
- if (ah->ah_setup_xtx_desc(ah, NULL, 0, 0, 0, 0, 0, 0))
+ ret = ah->ah_setup_xtx_desc(ah, NULL, 0, 0, 0, 0, 0, 0);
+ if (ret < 0)
+ goto err;
+ if (ret > 0)
__set_bit(ATH_STAT_MRRETRY, sc->status);

/*
diff --git a/drivers/net/wireless/ath5k/hw.c b/drivers/net/wireless/ath5k/hw.c
index 93a75f2..463413a 100644
--- a/drivers/net/wireless/ath5k/hw.c
+++ b/drivers/net/wireless/ath5k/hw.c
@@ -45,7 +45,7 @@ static int ath5k_hw_setup_4word_tx_desc(struct ath5k_hw *, struct ath5k_desc *,
unsigned int, unsigned int, enum ath5k_pkt_type, unsigned int,
unsigned int, unsigned int, unsigned int, unsigned int, unsigned int,
unsigned int, unsigned int);
-static bool ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *, struct ath5k_desc *,
+static int ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *, struct ath5k_desc *,
unsigned int, unsigned int, unsigned int, unsigned int, unsigned int,
unsigned int);
static int ath5k_hw_proc_4word_tx_status(struct ath5k_hw *, struct ath5k_desc *);
@@ -3733,7 +3733,7 @@ static int ath5k_hw_setup_4word_tx_desc(struct ath5k_hw *ah,
/*
* Initialize a 4-word multirate tx descriptor on 5212
*/
-static bool
+static int
ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
unsigned int tx_rate1, u_int tx_tries1, u_int tx_rate2, u_int tx_tries2,
unsigned int tx_rate3, u_int tx_tries3)
@@ -3773,10 +3773,10 @@ ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,

#undef _XTX_TRIES

- return true;
+ return 1;
}

- return false;
+ return 0;
}

/*
--
1.5.3.8

2008-02-15 21:01:19

by Jiri Slaby

[permalink] [raw]
Subject: [PATCH 3/3] WDEV: ath5k, typecheck on nonDEBUG

At least type check the ATH5K_TRACE paramter on !ATH5K_DEBUG configs.

Signed-off-by: Jiri Slaby <[email protected]>
Cc: Nick Kossifidis <[email protected]>
Cc: Luis R. Rodriguez <[email protected]>
---
drivers/net/wireless/ath5k/debug.h | 4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/drivers/net/wireless/ath5k/debug.h b/drivers/net/wireless/ath5k/debug.h
index c4fd8c4..44324c3 100644
--- a/drivers/net/wireless/ath5k/debug.h
+++ b/drivers/net/wireless/ath5k/debug.h
@@ -171,7 +171,9 @@ ath5k_debug_printtxbuf(struct ath5k_softc *sc,

#else /* no debugging */

-#define ATH5K_TRACE(_sc) /* empty */
+#include <linux/compiler.h>
+
+#define ATH5K_TRACE(_sc) typecheck(struct ath5k_softc *, (_sc))

static inline void __attribute__ ((format (printf, 3, 4)))
ATH5K_DBG(struct ath5k_softc *sc, unsigned int m, const char *fmt, ...) {}
--
1.5.3.8

2008-02-15 22:00:51

by Nick Kossifidis

[permalink] [raw]
Subject: Re: [PATCH 1/3] WDEV: ath5k, fix lock imbalance

2008/2/15, Jiri Slaby <[email protected]>:
> Omitted lock causes sparse warning
> drivers/net/wireless/ath5k/base.c:1682:1: warning: context imbalance in 'ath5k_tasklet_rx' - different lock contexts for basic block
>
> Add the lock to the guilty fail path.
>
> Signed-off-by: Jiri Slaby <[email protected]>
> Cc: Nick Kossifidis <[email protected]>
> Cc: Luis R. Rodriguez <[email protected]>
> ---
> drivers/net/wireless/ath5k/base.c | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/net/wireless/ath5k/base.c b/drivers/net/wireless/ath5k/base.c
> index ddc8714..49d38e8 100644
> --- a/drivers/net/wireless/ath5k/base.c
> +++ b/drivers/net/wireless/ath5k/base.c
> @@ -1715,6 +1715,7 @@ ath5k_tasklet_rx(unsigned long data)
> break;
> else if (unlikely(ret)) {
> ATH5K_ERR(sc, "error in processing rx descriptor\n");
> + spin_unlock(&sc->rxbuflock);
> return;
> }
>
>


Acked-by: Nick Kossifidis <[email protected]>


--
GPG ID: 0xD21DB2DB
As you read this post global entropy rises. Have Fun ;-)
Nick

2008-02-15 22:03:24

by Nick Kossifidis

[permalink] [raw]
Subject: Re: [PATCH 2/3] WDEV, ath5k, don't return int from bool function

2008/2/15, Jiri Slaby <[email protected]>:
> sparse sees int -> bool cast as an error:
> hw.c:3754:10: warning: cast truncates bits from constant value (ffffffea becomes 0)
> Fix it by converting the rettype to int and check appropriately.
>
> Signed-off-by: Jiri Slaby <[email protected]>
> Cc: Nick Kossifidis <[email protected]>
> Cc: Luis R. Rodriguez <[email protected]>
> ---
> drivers/net/wireless/ath5k/ath5k.h | 2 +-
> drivers/net/wireless/ath5k/base.c | 5 ++++-
> drivers/net/wireless/ath5k/hw.c | 8 ++++----
> 3 files changed, 9 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/net/wireless/ath5k/ath5k.h b/drivers/net/wireless/ath5k/ath5k.h
> index c79066b..69dea33 100644
> --- a/drivers/net/wireless/ath5k/ath5k.h
> +++ b/drivers/net/wireless/ath5k/ath5k.h
> @@ -1035,7 +1035,7 @@ struct ath5k_hw {
> unsigned int, unsigned int, enum ath5k_pkt_type, unsigned int,
> unsigned int, unsigned int, unsigned int, unsigned int,
> unsigned int, unsigned int, unsigned int);
> - bool (*ah_setup_xtx_desc)(struct ath5k_hw *, struct ath5k_desc *,
> + int (*ah_setup_xtx_desc)(struct ath5k_hw *, struct ath5k_desc *,
> unsigned int, unsigned int, unsigned int, unsigned int,
> unsigned int, unsigned int);
> int (*ah_proc_tx_desc)(struct ath5k_hw *, struct ath5k_desc *);
> diff --git a/drivers/net/wireless/ath5k/base.c b/drivers/net/wireless/ath5k/base.c
> index 49d38e8..59e5d56 100644
> --- a/drivers/net/wireless/ath5k/base.c
> +++ b/drivers/net/wireless/ath5k/base.c
> @@ -668,7 +668,10 @@ ath5k_attach(struct pci_dev *pdev, struct ieee80211_hw *hw)
> * return false w/o doing anything. MAC's that do
> * support it will return true w/o doing anything.
> */
> - if (ah->ah_setup_xtx_desc(ah, NULL, 0, 0, 0, 0, 0, 0))
> + ret = ah->ah_setup_xtx_desc(ah, NULL, 0, 0, 0, 0, 0, 0);
> + if (ret < 0)
> + goto err;
> + if (ret > 0)
> __set_bit(ATH_STAT_MRRETRY, sc->status);
>
> /*
> diff --git a/drivers/net/wireless/ath5k/hw.c b/drivers/net/wireless/ath5k/hw.c
> index 93a75f2..463413a 100644
> --- a/drivers/net/wireless/ath5k/hw.c
> +++ b/drivers/net/wireless/ath5k/hw.c
> @@ -45,7 +45,7 @@ static int ath5k_hw_setup_4word_tx_desc(struct ath5k_hw *, struct ath5k_desc *,
> unsigned int, unsigned int, enum ath5k_pkt_type, unsigned int,
> unsigned int, unsigned int, unsigned int, unsigned int, unsigned int,
> unsigned int, unsigned int);
> -static bool ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *, struct ath5k_desc *,
> +static int ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *, struct ath5k_desc *,
> unsigned int, unsigned int, unsigned int, unsigned int, unsigned int,
> unsigned int);
> static int ath5k_hw_proc_4word_tx_status(struct ath5k_hw *, struct ath5k_desc *);
> @@ -3733,7 +3733,7 @@ static int ath5k_hw_setup_4word_tx_desc(struct ath5k_hw *ah,
> /*
> * Initialize a 4-word multirate tx descriptor on 5212
> */
> -static bool
> +static int
> ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
> unsigned int tx_rate1, u_int tx_tries1, u_int tx_rate2, u_int tx_tries2,
> unsigned int tx_rate3, u_int tx_tries3)
> @@ -3773,10 +3773,10 @@ ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
>
> #undef _XTX_TRIES
>
> - return true;
> + return 1;
> }
>
> - return false;
> + return 0;
> }
>
> /*
>


Acked-by: Nick Kossifidis <[email protected]>


--
GPG ID: 0xD21DB2DB
As you read this post global entropy rises. Have Fun ;-)
Nick

2008-02-15 22:08:19

by Luis R. Rodriguez

[permalink] [raw]
Subject: Re: [PATCH 2/3] WDEV, ath5k, don't return int from bool function

On Fri, Feb 15, 2008 at 3:58 PM, Jiri Slaby <[email protected]> wrote:

> -static bool
> +static int
> ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
> unsigned int tx_rate1, u_int tx_tries1, u_int tx_rate2, u_int tx_tries2,
> unsigned int tx_rate3, u_int tx_tries3)
> @@ -3773,10 +3773,10 @@ ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
>
> #undef _XTX_TRIES
>
> - return true;
> + return 1;
> }
>
> - return false;
> + return 0;
> }

Shouldn't we then treat 0 as OK?

Luis

2008-02-15 22:34:54

by Luis R. Rodriguez

[permalink] [raw]
Subject: Re: [PATCH 3/3] WDEV: ath5k, typecheck on nonDEBUG

On Fri, Feb 15, 2008 at 3:58 PM, Jiri Slaby <[email protected]> wrote:
> At least type check the ATH5K_TRACE paramter on !ATH5K_DEBUG configs.

That's pretty cool.

> Signed-off-by: Jiri Slaby <[email protected]>
> Cc: Nick Kossifidis <[email protected]>
> Cc: Luis R. Rodriguez <[email protected]>

Acked-by: Luis R. Rodriguez <[email protected]>

Luis

2008-02-15 22:54:25

by Jiri Slaby

[permalink] [raw]
Subject: Re: [PATCH 2/3] WDEV, ath5k, don't return int from bool function

On 02/15/2008 11:08 PM, Luis R. Rodriguez wrote:
> On Fri, Feb 15, 2008 at 3:58 PM, Jiri Slaby <[email protected]> wrote:
>
>> -static bool
>> +static int
>> ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
>> unsigned int tx_rate1, u_int tx_tries1, u_int tx_rate2, u_int tx_tries2,
>> unsigned int tx_rate3, u_int tx_tries3)
>> @@ -3773,10 +3773,10 @@ ath5k_hw_setup_xr_tx_desc(struct ath5k_hw *ah, struct ath5k_desc *desc,
>>
>> #undef _XTX_TRIES
>>
>> - return true;
>> + return 1;
>> }
>>
>> - return false;
>> + return 0;
>> }
>
> Shouldn't we then treat 0 as OK?

Sorry, I don't understand you. There is return -EINVAL in the function above
this too and we need to cope with another two states but the error: it is
supported/it isn't. You mean to consider 0 as supported, -ENODEV/-EOPNOTSUPP as
unsupported and the rest as error?

2008-02-15 23:24:59

by Luis R. Rodriguez

[permalink] [raw]
Subject: Re: [PATCH 2/3] WDEV, ath5k, don't return int from bool function

On Fri, Feb 15, 2008 at 5:38 PM, Jiri Slaby <[email protected]> wrote:
>
> You mean to consider 0 as supported, -ENODEV/-EOPNOTSUPP as
> unsupported and the rest as error?
>

Yeap, what do you think?

Luis

2008-02-15 23:29:49

by Jiri Slaby

[permalink] [raw]
Subject: Re: [PATCH 2/3] WDEV, ath5k, don't return int from bool function

On 02/16/2008 12:24 AM, Luis R. Rodriguez wrote:
> On Fri, Feb 15, 2008 at 5:38 PM, Jiri Slaby <[email protected]> wrote:
>> You mean to consider 0 as supported, -ENODEV/-EOPNOTSUPP as
>> unsupported and the rest as error?
>>
>
> Yeap, what do you think?

I have no opinion in this case, which one do you (all) like more?