#482: spin_lock_irqsave problem
---------------------------------+------------------------------------------
Reporter: anonymous | Owner:
Type: defect | Status: new
Priority: major | Milestone:
Component: madwifi: other | Version:
Resolution: | Keywords:
Patch_attached: 0 |
---------------------------------+------------------------------------------
Old description:
> as i remember you must use only local variable to save flags on funcs
> like save_flags(), spin_lock_irqsave() etc. (maybe i'm wrong?). Now look
> at current code in madwifi_ng:
> #define ATH_NODE_UAPSD_LOCK_IRQ(_an)
> spin_lock_irqsave(&(_an)->an_uapsd_lock, (_an)->an_uapsd_lockflags)
> #define ATH_TXQ_LOCK_IRQ(_tq)
> spin_lock_irqsave(&(_tq)->axq_lock, (_tq)->axq_lockflags)
> #define ATH_RXBUF_LOCK_IRQ(_sc)
> spin_lock_irqsave(&(_sc)->sc_rxbuflock, (_sc)->sc_rxbuflockflags)
> it seems to me that you need macros like that:
> #define ATH_RXBUF_LOCK_IRQSAVE(_sc) do { \
> unsigned long __ilockflags; \
> spin_lock_irqsave(&(_sc)->sc_rxbuflock, __ilockflags);
> #define ATH_RXBUF_UNLOCK_IRQRESTORE(_sc) \
> spin_unlock_irqrestore(&(_sc)->sc_rxbuflock, __ilockflags); \
> } while (0)
New description:
as i remember you must use only local variable to save flags on funcs like
save_flags(), spin_lock_irqsave() etc. (maybe i'm wrong?). Now look at
current code in madwifi_ng:
{{{
#define ATH_NODE_UAPSD_LOCK_IRQ(_an)
spin_lock_irqsave(&(_an)->an_uapsd_lock, (_an)->an_uapsd_lockflags)
#define ATH_TXQ_LOCK_IRQ(_tq)
spin_lock_irqsave(&(_tq)->axq_lock, (_tq)->axq_lockflags)
#define ATH_RXBUF_LOCK_IRQ(_sc)
spin_lock_irqsave(&(_sc)->sc_rxbuflock, (_sc)->sc_rxbuflockflags)
}}}
it seems to me that you need macros like that:
{{{
#define ATH_RXBUF_LOCK_IRQSAVE(_sc) do { \
unsigned long __ilockflags; \
spin_lock_irqsave(&(_sc)->sc_rxbuflock, __ilockflags);
#define ATH_RXBUF_UNLOCK_IRQRESTORE(_sc) \
spin_unlock_irqrestore(&(_sc)->sc_rxbuflock, __ilockflags); \
} while (0)
}}}
Comment (by mrenzmann):
Nice attitude... "I'm too lazy to learn the syntax, so I submit a mostly
unreadable ticket and let someone else handle that stuff". This is exactly
the kind of attitude that makes a project like this one going. Thanks. :-(
In addition the subject of your ticket isn't really a matter for a ticket,
but would be better suited for a message to madwifi-devel.
--
Ticket URL: <http://madwifi.org/ticket/482>
MadWifi <http://madwifi.org/>
Multiband Atheros Driver for Wireless Fidelity
