On Fri, Dec 14, 2001 at 07:01:23PM +0530, V Ganesh wrote: > hi, > I've been developing a usb driver for the ipaq running wince. I ran > into some trouble which, I think, potentially affects the visor driver as well. > visor_write() calls kmalloc(..., GFP_KERNEL). I did the same thing and got > hit by a BUG() in kmem_cache_grow(). > if (in_interrupt() && (flags & SLAB_LEVEL_MASK) != SLAB_ATOMIC) > BUG(); > I think the problem is that the line discipline (ppp in my case) can call > visor_write() in bottom-half context. if you're lucky and don't have to grow a > slab for your kmalloc(), you won't hit it.
Which kernel version of the visor driver are you looking at? The most recent one does not call kmalloc in the write() call (well ok, the call is still there, but it never gets executed, as the urb transfer buffers all get set up in the visor_init() function. I should just remove it, but I like to be safe :) > now that the ipaq driver works reasonably well, I'd appreciate some > pointers on how to get it tested by a wider audience and any other things which > need to be done before submitting to the maintainer. should I just mail it to > the list? Mail it to the list, and copy the subsystem maintainer. That should get you the widest audience that cares about these things. thanks, greg k-h _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
