RE: bug in sscanf()?
From: Allan, Bruce W
Date: Mon Jan 13 2014 - 19:32:22 EST
> -----Original Message-----
> From: linus971@xxxxxxxxx [mailto:linus971@xxxxxxxxx] On Behalf Of Linus
> Torvalds
> Sent: Monday, January 13, 2014 4:23 PM
> To: Al Viro
> Cc: Allan, Bruce W; linux-kernel@xxxxxxxxxxxxxxx; Jan Beulich; Alexey
> Dobriyan
> Subject: Re: bug in sscanf()?
>
> On Tue, Jan 14, 2014 at 6:30 AM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > Comments?
>
> Do we have actual users of this? Because I'd almost be inclined to say
> "we just don't support field widths on sscanf() and will warn" unless
> there are users.
>
> We've done that before. The kernel has various limited functions. See
> the whole snprint() issue with %n, which we decided that supporting
> the full semantics was actually a big mistake and we actively
> *removed* code that had been misguidedly added just because people
> thought we should do everything a standard user library does..
>
> Limiting our problem space is a *good* thing, not a bad thing.
>
> If it's possible, of course, and we don't have nasty users.
>
> Linus
I was hoping to use sscanf() in this way for a driver I'm working on to support
Thunderbolt device authentication, but if it's too much to ask for I could probably
work around this.
Bruce.
èº{.nÇ+·®+%Ëlzwm
ébëæìr¸zX§»®w¥{ayºÊÚë,j¢f£¢·hàz¹®w¥¢¸¢·¦j:+v¨wèjØm¶ÿ¾«êçzZ+ùÝj"ú!¶iOæ¬z·vØ^¶m§ÿðÃnÆàþY&