Re: [avr-libc-dev] [bugs #10420] sscanf(), If there is a decimal number followed by the character '_' in the buffer, you can't get the the number.

[Joerg Wunsch]

As Joerg Wunsch wrote:

> I think I see the logical flaw in this now, too.  I tried to minimize
> the code required for this check, but obviously opened a window for
> false positives by that.

I tried to really understand what was going on, and debugged it a bit
further.  I can now clearly see the problem, and have another
suggested fix (appended).  Not only the code is documented better now,
but I'm also saving a few bytes compared to Ted's version.

My version:

% avr-size {.,avr3/,avr4/,avr5/}/build/libc/stdio/vfscanf_std.o
   text    data     bss     dec     hex filename
   1370       0       0    1370     55a ./build/libc/stdio/vfscanf_std.o
   1502       0       0    1502     5de avr3//build/libc/stdio/vfscanf_std.o
   1256       0       0    1256     4e8 avr4//build/libc/stdio/vfscanf_std.o
   1390       0       0    1390     56e avr5//build/libc/stdio/vfscanf_std.o

Ted's version:

% avr-size {.,avr3/,avr4/,avr5/}/build/libc/stdio/vfscanf_std.o
   text    data     bss     dec     hex filename
   1378       0       0    1378     562 ./build/libc/stdio/vfscanf_std.o
   1514       0       0    1514     5ea avr3//build/libc/stdio/vfscanf_std.o
   1260       0       0    1260     4ec avr4//build/libc/stdio/vfscanf_std.o
   1396       0       0    1396     574 avr5//build/libc/stdio/vfscanf_std.o

In comparision, the wrong avr2 version (before the patch) was 1366
bytes long.

If nobody sees a problem with that patch, I'd like to commit it.

-- 
J"org Wunsch                                           Unix support engineer
Wir stellen aus! Auf der SYSTEMS 2004  vom 18.-22. Oktober in München
Halle B 3, Stand 320-206 (Partner-Stand von Sun Microsystems)
Halle B 2, Stand 605 "Die Musterfirma"

scanf.diff
Description: Text document