On Fri, 6 Feb 2004, David Brownell wrote: > Date: Fri, 06 Feb 2004 11:57:41 -0800 > From: David Brownell <[EMAIL PROTECTED]> > To: [EMAIL PROTECTED] > Cc: linux-usb-devel <[EMAIL PROTECTED]>, > John Heil <[EMAIL PROTECTED]> > Subject: Re: [linux-usb-devel] Re: Test of ehci iso patch > > Michal Sojka wrote: > > > *frames per urb > > * *lengths of valid data > > * *length of mess > > * > > 16 > > 27520 > > 4472 > > But you said this was at 2 KB/uframe high bandwidth rate, so that > math doesn't check out: 16 frames/urb * 8 uframes/frame * 2 KB/uframe > means 256 KByte/urb, not 128 bytes shy of 27 KB. And if you meant > 16 uframes per urb, 27 KB takes almost 14 uframes, not 16. > > > > Because patterns of valid and invalid data are very regular, I think > > that it may be a bug in itd sheduling. > > Or maybe the buffer overflow (into next page) logic filling up the ITDs, > given the other cases were exactly one or two packets. You might have > a look at that part of the code. Or maybe you can just post a way for > someone else to reproduce the problem ... :)
Hi folks, Just a quick FYI on a buffer overflow I hit... In debugging my version, I hit a case of usb_alloc_urb where I used GFP_KERNEL. The request happened to be fullfilled at the end of a page so the some of the URB would cross the page boundary. The kmalloc did _not_ allocate the requested amount amount that was acrossed the page boundary, bug did not tell me the kmalloc failed... oops. Not having the time to debug kmalloc, I changed the GFP_KERNEL to SLAB_KERNEL and that solved it :). Cheers, johnh > > - Dave > > > - ----------------------------------------------------------------- John Heil South Coast Software Custom systems software for UNIX and IBM MVS mainframes 1-714-774-6952 [EMAIL PROTECTED] http://www.sc-software.com ----------------------------------------------------------------- ------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
