On Wed, 5 Jan 2011, Pete Zaitcev wrote: > On Wed, 5 Jan 2011 11:42:51 -0500 (EST) > Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > > > One aspect of this worries me: By copying the entire buffer, we risk > > leaking kernel data. What do you think? Should mon_submit() > > zero out the buffer for isochronous-IN transfers? > > I'm surprised that an access to keystrokes and/or filesystem data does > not bother you, but some random kernel data is a problem. > > I'll look at the patch, thanks. Probably the previous one was incomplete. This is what I get for not using git. Looking back through the email archive, I found the patch you submitted back in November to fix this same bug. The thing is, your patch rearranges the data by skipping parts that weren't received. This means isochronous-OUT submissions and isochronous-IN completions have to be parsed differently, which is bad. Also, you didn't make a corresponding change to the mon_text interface, and the data format change wasn't documented. On the whole, I don't think it's bad to copy the unused buffer contents. You do the same thing now for isochronous-OUT submissions (the packets need not be contiguous in the buffer, and you don't snip out any unused bytes). Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
