[ENH] PNGReadWriter writing support & alpha fixes (v3)
John M McIntosh
johnmci at smalltalkconsulting.com
Fri Feb 13 02:01:37 UTC 2004
The failure is in
deflateBlockchainLengthgoodMatch
(gdb) x/8i deflateBlockchainLengthgoodMatch+0x268
0x57d2c <deflateBlockchainLengthgoodMatch+616>: lwzx r29,r19,r3
0x57d30 <deflateBlockchainLengthgoodMatch+620>: addi r10,r29,1
0x57d34 <deflateBlockchainLengthgoodMatch+624>: stwx r10,r19,r3
0x57d38 <deflateBlockchainLengthgoodMatch+628>: lwz r8,0(r4)
0x57d3c <deflateBlockchainLengthgoodMatch+632>: lwz r7,0(r12)
0x57d40 <deflateBlockchainLengthgoodMatch+636>: lwz r6,0(r17)
r3: 0x05d765b4
r19: 0xcccccccc
Yes that would cause a fault...
L20:
.stabd 68,0,289
lwz r22,76(r1)
slwi r19,r2,2
.stabd 68,0,290
addis r11,r31,ha16(_zipLiteralCount-"L00000000038$pb")
la r4,lo16(_zipLiteralCount-"L00000000038$pb")(r11)
.stabd 68,0,292
addis r2,r31,ha16(_zipLiteralSize-"L00000000038$pb")
li r25,0
la r12,lo16(_zipLiteralSize-"L00000000038$pb")(r2)
.stabd 68,0,289
lwz r3,0(r22)
lwzx r29,r19,r3 <======= CRASH ============= I think it's the
((zipDistanceFreq[distance]) + 1);
addi r10,r29,1
stwx r10,r19,r3
.stabd 68,0,290
lwz r8,0(r4)
.stabd 68,0,292
lwz r7,0(r12)
.stabd 68,0,291
lwz r6,0(r17)
.stabd 68,0,290
addi r9,r8,1
.stabd 68,0,292
cmpw cr6,r9,r7
.stabd 68,0,290
stw r9,0(r4)
.stabd 68,0,291
addi r5,r6,1
stw r5,0(r17)
.stabd 68,0,292
Line 289
zipDistanceFreq[distance] = ((zipDistanceFreq[distance]) + 1);
zipLiteralCount += 1;
zipMatchCount += 1;
flushNeeded = (zipLiteralCount == zipLiteralSize) ||
(((zipLiteralCount & 4095) == 0) && (shouldFlush()));
On Feb 12, 2004, at 5:35 PM, Bert Freudenberg wrote:
>
> Am 12.02.2004 um 23:03 schrieb Andreas Raab:
>
>> This is now v3 which fixes a bad, bad CRC issue, adds CRC validation
>> and
>> hopefully fixes the byte-sex problems.
>>
>> Avi - if you give this a try and the tests fail, can you send me the
>> generated test images (they will be written out automatically).
>> Off-list
>> please ;-)
>
> Oh, it's worse than before - it crashes the MacVM (log attached for
> John)! After commenting out test8BitDisplay the tests run with this
> result: 24 run, 23 passed, 0 failed, 1 error. The error is in
> test4BitDisplay (error out of bounds in ZipEncoder. But the tests seem
> to be wrong too, as none of the generated test images can be loaded by
> another application (Preview in my case). I'll send the images
> off-line.
>
> - Bert -
>
> <Squeak VM Opt.crash.log.gz>
>
--
========================================================================
===
John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com
========================================================================
===
More information about the Squeak-dev
mailing list
|