medcon recompiled without optimization and with debugging symbols:
CFLAGS="-g -O0" DEB_BUILD_OPTIONS="debug, noopt, nostrip" dpkg-buildpackage -uc
-us -rfakeroot
Running on Debian Unstable AMD64.
Program received signal SIGSEGV, Segmentation fault.
0x00002aea4eac2d45 in DecodeFirstRow (dcPtr=0x2aea4ec09de0,
curRowBuf=0x2aea4ec09ec0) at huffd.c:472
472 curRowBuf[col][curComp]=d+curRowBuf[col-1][curComp];
(gdb) bt
#0 0x00002aea4eac2d45 in DecodeFirstRow (dcPtr=0x2aea4ec09de0,
curRowBuf=0x2aea4ec09ec0) at huffd.c:472
#1 0x00002aea4eac2edc in DecodeImage (dcPtr=0x2aea4ec09de0,
image=0x7fff5c29e030, depth=16) at huffd.c:536
#2 0x00002aea4eac21c2 in JPEGLosslessDecodeImage (inFile=0x516010,
image16=0x2aea4f245010, depth=16, length=5321384) at decomp.c:90
#3 0x00002aea4eac1fe8 in mdc_dicom_decomp_ljpg (fp=0x516010,
image16=0x2aea4f245010, length=5321384, depth=16) at decomp.c:177
#4 0x00002aea4eabdafb in mdc_dicom_decompress (s=0x2aea4ec09d60,
e=0x2aea4ec09d00) at basic.c:1035
#5 0x00002aea4eabe6aa in dicom_single () at single.c:288
#6 0x00002aea4eac019f in dicom_read (file=0x7fff5c29e310 "20061119171007906",
image=0x7fff5c29e248, images=0x7fff5c29e244, parametric=1) at image.c:83
#7 0x00002aea4eab96a9 in mdc_dicom_read (fi=0x7fff5c29e2f0,
image=0x7fff5c29e248, number=0x7fff5c29e244) at m-dicm.c:2146
#8 0x00002aea4eab2684 in MdcReadDICM (fi=0x7fff5c29e2f0) at m-dicm.c:485
#9 0x00002aea4ea70bf5 in MdcReadFile (fi=0x7fff5c29e2f0, filenr=1, ReadFunc=0)
at m-files.c:192
#10 0x0000000000401475 in main (argc=3, argv=0x7fff5c29ec98) at medcon.c:96
(gdb) print d
$1 = 13
(gdb) print curComp
$2 = 0
(gdb) print col
$3 = 2048
Notice the very round and nice value of "col"? Looks suspicious....
How big is the curRowBuf anyway?
./libs/ljpg/huffd.c -- DecodeImage():
curRowBuf = mcuROW1;
./libs/ljpg/jpegutil.c:
static char mcuROW1Memory[1024 * sizeof(MCU)];
./libs/ljpg/mcu.h:
typedef unsigned short ComponentType; /* the type of image components */
typedef ComponentType *MCU; /* MCU - array of samples */
test.c tell me that sizeof(unsigned short) is 2, so curRowBuf is 2048.
Should the row be dynamic length instead of "1024 * sizeof(MCU)" ?
Should col never reach that high value?
--
Regards,
Andreas Henriksson
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
