Interesting failures of 'cscope'

Interesting failures of 'cscope'

[Carl Spalletta]

FYI

  The first failure seems to be a confusion between function declarations and definitions - 
eg for linux-2.6.7, it says FsmNew() is called by drivers/isdn/hisax/fsm.h::CallcNew, thus:

$cscope -d -p9 -L -3 FsmNew
  ...
  drivers/isdn/hisax/fsm.h CallcNew 50 int FsmNew(struct Fsm *fsm, struct FsmNode,*fnlist, int
fncount);
  ...

  but there is no call there, only an external declaration of FsmNew, and no declaration of
CallcNew of any kind whatsoever in that file!


  The second failure appears to be related to an inability to cope with complicated declarations
within a prototype, such of those of type pointer-to-function. For example, the definition:

  struct net_device *alloc_netdev(int sizeof_priv, const char *mask, 
                                         void (*setup)(struct net_device *))

from line 73 of drivers/net/net_init.c, is not recognized by cscope as a valid function definition
and so it does not find the call to kmalloc (or anything else) contained in that function.

$ cscope -d -p9 -L -1 alloc_netdev
<nothing>
$ cscope -d -p9 -L -3 kmalloc | grep alloc_netdev
<nothing>

Re: Interesting failures of 'cscope'

[Martin J. Bligh]

Various versions of cscope are buggy as hell, and I'm not at all sure the
recent ones are any better than older ones ... caveat user ;-)

Personally I use an old binary ripped off RH 6.2, which still seems to
work ... any current ones I've tried are broken.

M.

--On Wednesday, August 04, 2004 16:22:52 -0700 Carl Spalletta <cspalletta@yahoo.com> wrote:

> FYI
> 
>   The first failure seems to be a confusion between function declarations and definitions - 
> eg for linux-2.6.7, it says FsmNew() is called by drivers/isdn/hisax/fsm.h::CallcNew, thus:
> 
> $cscope -d -p9 -L -3 FsmNew
>   ...
>   drivers/isdn/hisax/fsm.h CallcNew 50 int FsmNew(struct Fsm *fsm, struct FsmNode,*fnlist, int
> fncount);
>   ...
> 
>   but there is no call there, only an external declaration of FsmNew, and no declaration of
> CallcNew of any kind whatsoever in that file!
> 
> 
>   The second failure appears to be related to an inability to cope with complicated declarations
> within a prototype, such of those of type pointer-to-function. For example, the definition:
> 
>   struct net_device *alloc_netdev(int sizeof_priv, const char *mask, 
>                                          void (*setup)(struct net_device *))
> 
> from line 73 of drivers/net/net_init.c, is not recognized by cscope as a valid function definition
> and so it does not find the call to kmalloc (or anything else) contained in that function.
> 
> $ cscope -d -p9 -L -1 alloc_netdev
> <nothing>
> $ cscope -d -p9 -L -3 kmalloc | grep alloc_netdev
> <nothing>
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
> 

Re: Interesting failures of 'cscope'

[Carl Spalletta]

The previous posting related to false negatives obtained by 'cscope'. This posting relates
to false positives from the same cause (can't parse pointer-to-function declarations):

>From arch/i386/kernel/irq.c:

    621 int request_irq(unsigned int irq,
    622                 irqreturn_t (*handler)(int, void *, struct pt_regs *),
    623                 unsigned long irqflags,
    624                 const char * devname,
    625                 void *dev_id)

$ /usr/bin/cscope -d -p9 -L -3 setup_irq
...
arch/i386/kernel/irq.c irqreturn_t 660 retval = setup_irq(irq, action);
...

There is obviously no function called 'irqreturn_t'.