claunia/libcdio-osx
Archived
1
0
Fork 0
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

Work around problem where we were accessing outside of an allocate

Browse Source 
range when the drive endian was different than the CD-ROM
endianness. We do this by always allocating an extra block, but it
would be better to understand whether this is correct or whether some
logic needs to be fixed.
This commit is contained in:
rocky
2005-01-23 14:05:19 +00:00
parent 410a6a70e8
commit bdf0d1a6c4
2 changed files with 37 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
lib/cdda_interface/interface.c
View File

@@ -1,5 +1,5 @@
/*
$Id: interface.c,v 1.14 2005/01/18 16:05:29 rocky Exp $
$Id: interface.c,v 1.15 2005/01/23 14:05:19 rocky Exp $
Copyright (C) 2005 Rocky Bernstein <rocky@panix.com>
Copyright (C) 1998 Monty xiphmont@mit.edu
@@ -132,6 +132,13 @@ long cdda_read(cdrom_drive_t *d, void *buffer, lsn_t beginsector, long sectors)
uint16_t *p=(uint16_t *)buffer;
long els=sectors*CDIO_CD_FRAMESIZE_RAW/2;
/* Note: Something perhaps in the original cdparanioa code might
cause the code to access outside of the allocated range of
buffer. This comment is just to serve as a marker for
the loop where the data got clobbered. I don't think this
code however is wrong. See the comment in i_read_c_block
of paranioa.c
*/
for(i=0;i<els;i++)
p[i]=UINT16_SWAP_LE_BE_C(p[i]);
}

26
lib/paranoia/paranoia.c
View File

@@ -1,5 +1,5 @@
/*
$Id: paranoia.c,v 1.8 2005/01/23 05:31:03 rocky Exp $
$Id: paranoia.c,v 1.9 2005/01/23 14:05:19 rocky Exp $
Copyright (C) 2004, 2005 Rocky Bernstein <rocky@panix.com>
Copyright (C) 1998 Monty xiphmont@mit.edu
@@ -1211,7 +1211,19 @@ i_read_c_block(cdrom_paranoia_t *p,long beginword,long endword,
new=new_c_block(p);
}
buffer=calloc(1, totaltoread*CDIO_CD_FRAMESIZE_RAW);
/* FIXME: We allocate one extra frame more than what we should be
using. In cdda_read() when the CD-ROM endian-ness is different
from the machine endian-ness and we need to swap bytes, we seem
to be accessing outside of buffer in some (but not all)
cases. This is probably due to this routine calling cdda_read
with parameter that would have it access outside the allocated
range. Since I don't know how to fix, we'll do the harmless over
allocation.
On a sparc (Big Endian) Solaris with a little-endian CD-ROM. The
symptom would be that the calloc *after* the one below would fail.
*/
buffer=calloc((totaltoread+1)*CDIO_CD_FRAMESIZE_RAW, 1);
sofar=0;
firstread=-1;
@@ -1235,15 +1247,19 @@ i_read_c_block(cdrom_paranoia_t *p,long beginword,long endword,
if (secread>0){
if (firstread<0) firstread = adjread;
if ((thisread=cdda_read(p->d,buffer+sofar*CD_FRAMEWORDS,adjread,
secread))<secread){
/* See FIXME note above about a possibly erroneous call here. */
thisread = cdda_read(p->d, buffer+sofar*CD_FRAMEWORDS, adjread, secread);
if ( thisread < secread) {
if (thisread<0) thisread=0;
/* Uhhh... right. Make something up. But don't make us seek
backward! */
if(callback)(*callback)((adjread+thisread)*CD_FRAMEWORDS,PARANOIA_CB_READERR);
if(callback)
(*callback)((adjread+thisread)*CD_FRAMEWORDS, PARANOIA_CB_READERR);
memset(buffer+(sofar+thisread)*CD_FRAMEWORDS,0,
CDIO_CD_FRAMESIZE_RAW*(secread-thisread));
if(flags)memset(flags+(sofar+thisread)*CD_FRAMEWORDS,2,