Hi Tom, DOS386, >> As already pointed: >> http://sourceforge.net/support/tracker.php?aid=2901916 >> there is a CRITICAL BUG in FreeDOS kernel 2039 :-( > >> I can well reproduce it (not always, though) when COPY'ing files onto >> a FAT28 volume. There can be even multiple crossy bunches per multiple > (>>=2) files. Also the old files are getting corrupted, the last >> written file can (?) be intact. Now I upgraded to 2038, let's see >> whether it also has this bug, EDR-DOS definitely doesn't. > > just wondering: could this be an issue with JEMM386 and/or UIDE ? > is this reproducable with only loading HIMEM and kernel ? > > Eric: you changed something in the FAT code. is this also in 2038 or > only in 2039 ?
If you have the patience to load this slow-CGI-generated page: http://freedos.svn.sf.net/viewvc/freedos/kernel/trunk/?sortby=date&view=log then you will see that there were no changes outside SYS past August 22 and that kernel 2038 was released in mid-may. Since then, there were big changes, often by Bart, for example a big patch series to get the f-nodes out of quite a bit of our code. Luckily the history.txt file is also very decent so you can just download the source zips of 2038 and 2039 and read about almost all changes in detail, with bug and revision numbers. Changes and people since 2038: - Nov 13 Jeremy changed only SYS - Jul 30 / Aug 1 Jeremy SYS and build system - Jul 13 Jeremy mostly comments, printf - Jul 1 Jeremy for SYS and COUNTRY SYS - Jun 22 Jeremy build system - Jun 10 Jeremy build system, EXEFLAT - May 18 Jeremy build system / version tags - May 17 Jeremy tag kernel 2038 (SVN revision 1381) - May 31 Eric updated history.txt and other docs - all other changes since kernel 2038 are by Bart :-) http://freedos.svn.sf.net/viewvc/freedos/kernel/trunk/kernel/?sortby=date http://freedos.svn.sf.net/viewvc/freedos/kernel/trunk/hdr/?sortby=date Some interesting updates: - irqstack.asm Aug 22 - fatdir.c many changes (revisions 1389-1392, 1398, 1404, 1405, 1410, 1419, 1420, 1432, 1433, 1435, 1436, 1438, 1439, 1441, 1468, 1473) Fatdir might be related to your bug. - fatfs.c many changes (1389, 1391-1393, 1397-1405, 1408-1411, 1416-1421, 1432, 1433, 1435-1437, 1441, 1442, 1444, 1445, 1447, 1448, 1451, 1453, 1458, 1473, 1490) Fatfs is even more likely to be related to your bug. - prf.c (1394, 1482, 1491) - various header changes http://freedos.svn.sf.net/viewvc/freedos/kernel/trunk/kernel/fatdir.c?view=log&sortby=date http://freedos.svn.sf.net/viewvc/freedos/kernel/trunk/kernel/fatfs.c?sortby=date&view=log Depending on how hard it is for you to compile and try whether the bug (or maybe it is a regression) exists in a certain version, I would recommend a binary search through SVN revisions... Current is revision 1500. If you assume that 2038 is revision 1381, then you can compile revision 1440. If it has the bug, compile 1410. If not, compile 1470. Try and compile 1395, 1425, 1455 or 1485 depending on the outcome of your test, and so on :-). The 1st test reduces search range to 60 revisions, next to 30, next to 15, a few more and voila. Eric ------------------------------------------------------------------------------ Return on Information: Google Enterprise Search pays you back Get the facts. http://p.sf.net/sfu/google-dev2dev _______________________________________________ Freedos-kernel mailing list Freedos-kernel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-kernel
