Hi Marti, sorry to come with the lcms-1.12 output. Unfortunedly the behaviour still remaines lcms-1.13 valgrind-2.1.0: ==2880== Invalid read of size 1 ==2880== at 0x40022270: memcpy (mac_replace_strmem.c:270) ==2880== by 0x4052BEDE: cmsReadICCText (cmsio1.c:1638) ==2880== by 0x4052CC53: cmsTakeManufacturer (cmsio1.c:2180) ==2880== by 0x808B16B: cms_get_profile_info (cms.c:555) ==2880== Address 0x43709DED is 1 bytes after a block of size 16 alloc'd ==2880== at 0x4002ABF6: malloc (vg_replace_malloc.c:160) ==2880== by 0x4052A19A: InitTag (cmsio1.c:581) ==2880== by 0x4052F2BC: _cmsAddTextTag (cmsio1.c:3410) ==2880== by 0x4052F672: cmsAddTag (cmsio1.c:3504) ... ==2880== Invalid read of size 1 ==2880== at 0x40022277: memcpy (mac_replace_strmem.c:271) ==2880== by 0x4052BEDE: cmsReadICCText (cmsio1.c:1638) ==2880== by 0x4052CCC3: cmsTakeModel (cmsio1.c:2197) ==2880== by 0x808ACD8: cms_get_long_profile_info (cms.c:408) ==2880== Address 0x43709E2A is 0 bytes after a block of size 14 alloc'd ==2880== at 0x4002ABF6: malloc (vg_replace_malloc.c:160) ==2880== by 0x4052A19A: InitTag (cmsio1.c:581) ==2880== by 0x4052F2BC: _cmsAddTextTag (cmsio1.c:3410) ==2880== by 0x4052F672: cmsAddTag (cmsio1.c:3504) ... ==2880== Use of uninitialised value of size 8 ==2880== at 0x4052C52A: cmsReadICCMatrixRGB2XYZ (cmsio1.c:1889) ==2880== by 0x40537234: cmsBuildInputMatrixShaper (cmsxform.c:503) ==2880== by 0x4053816A: PickTransformRoutine (cmsxform.c:1050) ==2880== by 0x4053845B: cmsCreateProofingTransform (cmsxform.c:1266) .. more
in 1.14 it seems all remained the same for this special issue: ==15696== Invalid read of size 1 ==15696== at 0x40022269: memcpy (mac_replace_strmem.c:269) ==15696== by 0x4052C0FE: cmsReadICCText (cmsio1.c:1646) ==15696== by 0x4052CE83: cmsTakeManufacturer (cmsio1.c:2195) ==15696== by 0x808B16B: cms_get_profile_info (cms.c:555) ==15696== Address 0x4372EBF0 is 0 bytes after a block of size 16 alloc'd ==15696== at 0x4002ABF6: malloc (vg_replace_malloc.c:160) ==15696== by 0x4052A3BA: InitTag (cmsio1.c:589) ==15696== by 0x4052F4FC: _cmsAddTextTag (cmsio1.c:3427) ==15696== by 0x4052F8B2: cmsAddTag (cmsio1.c:3521) ==15696== Invalid read of size 1 ==15696== at 0x40022270: memcpy (mac_replace_strmem.c:270) ==15696== by 0x4052C0FE: cmsReadICCText (cmsio1.c:1646) ==15696== by 0x4052D195: cmsTakeProductDesc (cmsio1.c:2287) ==15696== by 0x808B17E: cms_get_profile_info (cms.c:556) ==15696== Address 0x4372EC71 is 3 bytes after a block of size 14 alloc'd ==15696== at 0x4002ABF6: malloc (vg_replace_malloc.c:160) ==15696== by 0x4052A3BA: InitTag (cmsio1.c:589) ==15696== by 0x4052F4FC: _cmsAddTextTag (cmsio1.c:3427) ==15696== by 0x4052F8B2: cmsAddTag (cmsio1.c:3521) ==15696== Use of uninitialised value of size 8 ==15696== at 0x4052C751: cmsReadICCMatrixRGB2XYZ (cmsio1.c:1897) ==15696== by 0x405374A4: cmsBuildInputMatrixShaper (cmsxform.c:503) ==15696== by 0x4053839A: PickTransformRoutine (cmsxform.c:1043) ==15696== by 0x4053868B: cmsCreateProofingTransform (cmsxform.c:1259) Sorry to be so resistent, but I cannot open an untagged image in CinePaint colormanaged without crashing the whole application by the current behaviour. Again, it is only related to on the fly profiles not the other parts of lcms. thanks Kai-Uwe Am 10.08.04, 10:01 -0000 schrieb [EMAIL PROTECTED]: > > > > Hi Kai-Uwe, > > These are errors fixed in 1.13, though. See for example ReadICCAscii, > which no longer exists in latest sources. There are, however, other > small bugs that I've fixed in my development sources. Perhaps it would > be a good idea to do the release of 1.14 as a maintenance one. I tend > to do releases each 5-6 months to minimize noise, and was planning > some new features for 1.14, but those can wait as well... > > Opinions? > > Regards, > Marti. > > > Kai-Uwe Behrmann <[EMAIL PROTECTED]> said: > > > Hi, > > > > while valgrind-ing in CinePaint, I have many messages about memory access > > errors. - Valgrind is the memory debugger of the KDE project and others. > > Anyway I dont know how to observe an certain memory block for read/write > > access. Hints are very welcome! > > > > The problem with lcms for me is (while debugging), the hidding of > > the internal data structures (which is desireable for the lcms API) lets > > me jump from one function to the next, not beeing able to find and fix the > > bugs. > > > > It seems to me there occures overwritings of tags, and maybe afterwards > > corrupted memory blocks with futher errors. This is mostly during profile > > creation. I have not such problems, while using lcms for conversion, which > > seems more robust. > > > > lcms-1.1[2,3] valgrind-2.1.0: > > ==11550== Invalid read of size 1 > > ==11550== at 0x40022269: memcpy (mac_replace_strmem.c:269) > > ==11550== by 0x4052A6EE: ReadICCAscii (cmsio1.c:1563) > > ==11550== by 0x4052B373: cmsTakeManufacturer (cmsio1.c:2055) > > ==11550== by 0x808B16B: cms_get_profile_info (cms.c:555) > > ==11550== Address 0x43729F48 is 0 bytes after a block of size 16 alloc'd > > ==11550== at 0x4002ABF6: malloc (vg_replace_malloc.c:160) > > ==11550== by 0x40528ACA: InitTag (cmsio1.c:576) > > ==11550== by 0x4052D1FC: _cmsAddTextTag (cmsio1.c:3076) > > ==11550== by 0x4052D572: cmsAddTag (cmsio1.c:3158) > > and > > ==11550== Invalid read of size 1 > > ==11550== at 0x40022270: memcpy (mac_replace_strmem.c:270) > > ==11550== by 0x4052A6EE: ReadICCAscii (cmsio1.c:1563) > > ==11550== by 0x4052B665: cmsTakeProductDesc (cmsio1.c:2144) > > ==11550== by 0x808B17E: cms_get_profile_info (cms.c:556) > > ==11550== Address 0x43729FC9 is 3 bytes after a block of size 14 alloc'd > > ==11550== at 0x4002ABF6: malloc (vg_replace_malloc.c:160) > > ==11550== by 0x40528ACA: InitTag (cmsio1.c:576) > > ==11550== by 0x4052D1FC: _cmsAddTextTag (cmsio1.c:3076) > > ==11550== by 0x4052D572: cmsAddTag (cmsio1.c:3158) > > > > ==11550== Use of uninitialised value of size 8 > > ==11550== at 0x4052ACCA: cmsReadICCMatrixRGB2XYZ (cmsio1.c:1789) > > ==11550== by 0x40534B04: cmsBuildInputMatrixShaper (cmsxform.c:460) > > ==11550== by 0x4053582D: PickTransformRoutine (cmsxform.c:946) > > ==11550== by 0x40535B0B: cmsCreateProofingTransform (cmsxform.c:1169) > > > > Would be glad to use lcms internal on the fly profile creation :), but > > need first to make it more stable. I have workarrounds for all of this, > > but they are not that convenient. > > > > kind regards > > > > Kai-Uwe Behrmann > > + imaging developer / panoramas > > + color management > > + email: [EMAIL PROTECTED] > > > > > > > > > > ------------------------------------------------------- > > SF.Net email is sponsored by Shop4tech.com-Lowest price on Blank Media > > 100pk Sonic DVD-R 4x for only $29 -100pk Sonic DVD+R for only $33 > > Save 50% off Retail on Ink & Toner - Free Shipping and Free Gift. > > http://www.shop4tech.com/z/Inkjet_Cartridges/9_108_r285 > > _______________________________________________ > > Lcms-user mailing list > > [EMAIL PROTECTED] > > https://lists.sourceforge.net/lists/listinfo/lcms-user > > > > > > -- > > > Mit freundlichen Grüßen Kai-Uwe Behrmann + Bildprogrammierung / Panoramen + Farbmanagement + email: [EMAIL PROTECTED] ------------------------------------------------------- SF.Net email is sponsored by Shop4tech.com-Lowest price on Blank Media 100pk Sonic DVD-R 4x for only $29 -100pk Sonic DVD+R for only $33 Save 50% off Retail on Ink & Toner - Free Shipping and Free Gift. http://www.shop4tech.com/z/Inkjet_Cartridges/9_108_r285 _______________________________________________ Lcms-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/lcms-user
