On Sun, Feb 27, 2005 at 08:25:04PM +0000, [EMAIL PROTECTED] wrote:
> I decided to tweak sparse to give warnings on sizeof(pointer), so that I could
> check for other cases like this. The tweak was a very crude hack that I'm not
> proud of, and I am still trying to make it more reliable.
>
> So far I found 2 interesting cases (in 2.6.11-rc5). I'm not sure they are
> bugs,
> but they sure look suspicious.
>
> 1: drivers/usb/storage/usb.c:788
>
> /*
> * Since this is a new device, we need to register a SCSI
> * host definition with the higher SCSI layers.
> */
> us->host = scsi_host_alloc(&usb_stor_host_template, sizeof(us));
> if (!us->host) {
> printk(KERN_WARNING USB_STORAGE
> "Unable to allocate the scsi host\n");
> return -EBUSY;
> }
>
> "us" is a "struct us_data *". It seems pretty weird that we're allocating
> something the size of a pointer, and then waste a pointer to keep the address
> where it is allocated. So maybe this should be:
>
> us->host = scsi_host_alloc(&usb_stor_host_template, sizeof(*us));
This is actually correct as-is. We're allocating a host, and asking for
the sizeof(pointer) in the 'extra storage' section. We then store the
pointer (not what it points to) in the extra storage section a few lines down.
Matt
--
Matthew Dharm Home: [EMAIL PROTECTED]
Maintainer, Linux USB Mass Storage Driver
S: Another stupid question?
G: There's no such thing as a stupid question, only stupid people.
-- Stef and Greg
User Friendly, 7/15/1998
pgpkrUwIFp8xT.pgp
Description: PGP signature
