Re: [Freedos-devel] bug in freedos kernel int 21.1c invalid drive handling?
> As people might call this to find out whether a > drive exists, FreeDOS should return the sectors/ > cluster as 255 as expected for invalid drives... People might do, but it isn't reliable: redirected drives might return al=FFh as well. > The bugfix would make sure AL=255 is returned > for int 21.1c calls for an invalid drive, but > for CX and DX and DS:BX I am not sure whether > it is better to change or not to change them? > What do the other DOSes or DOSish OSes return? NTVDM 5.10 MS-DOS returns al=FFh and all other unchanged for invalid drives (as expected). Setting ds:bx unconditionally is a DOS-C bug (in inthndlr.c) that should be fixed. FatGetDrvData() returns a NULL pointer for invalid drives. >> 4) Other diferencies are in CD/DVD roms and flashdisks. > > Yeah, what happens for CD/DVD is kind of random I guess... I guess it depends on *CDEX, because DOS calls Int2F.110C for redirected drives. Christian -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
Re: [Freedos-devel] bug in freedos kernel int 21.1c invalid drive handling?
Hi Laaca, thanks for checking! > ... FreeDOS, MSDOS 7.1, Windows98 and WindowsXP ... > existing 3,5 floppy disks all give the same (correct) results only > MSDOS gives wrong Total number of Sectors and Sectors per Cluster As this is MS Win9x DOS 7, the wrong number is expected... > 2) For existing hard disks all three give different value for > Total Nr. of Sectors FD: 65525, MS: 65534, 98: 65526, XP: 15747 I assume you mean total number of clusters? And I assume you measured on a FAT32 drive which actually has more than 65536 clusters which cannot be expressed as 16 bit number? How many clusters does that drive have, for example according to dosfsck -v X: or so? > 3) Non existing drives: difference is in Sectors per Cluster: > in FreeDOS: all output seems to be "undefined", in other OSes > is all "undefined" except Sectors per Custer which seems to be > always 255 (FFh) As people might call this to find out whether a drive exists, FreeDOS should return the sectors/ cluster as 255 as expected for invalid drives... Trying with the FreeDOS 1.0 kernel 2036, I get: AX, CX, DX, BP, SI, DI, ES, SS, SP unchanged DS:BX 0:0, so I would not say "undefined". The experiment was to fill values like 1234 in all the registers to see what you get back later. The bugfix would make sure AL=255 is returned for int 21.1c calls for an invalid drive, but for CX and DX and DS:BX I am not sure whether it is better to change or not to change them? What do the other DOSes or DOSish OSes return? > 4) Other diferencies are in CD/DVD roms and flashdisks. Yeah, what happens for CD/DVD is kind of random I guess... Eric -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
Re: [Freedos-devel] bug in freedos kernel int 21.1c invalid drive handling?
> 1) For existing 3,5 floppy disks all give the same (correct) results only > MSDOS gives wrong Total number of Sectors and Sectors per Cluster This bug of MS-DOS 7.10 is also documented in RBIL 61. BTW, I think the other registers should not be changed when al is set to FFh regardless of what MS-DOS does. Christian -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
Re: [Freedos-devel] bug in freedos kernel int 21.1c invalid drive handling?
Well, I did some testing in FreeDOS, MSDOS 7.1, Windows98 and WindowsXP This function works in all these environment different. F 1) For existing 3,5 floppy disks all give the same (correct) results only MSDOS gives wrong Total number of Sectors and Sectors per Cluster 2) For existing hard disks all three give different value for Total Number of Sectors (FD: 65525, MS: 65534, 98: 65526, XP: 15747 (bullshit)) 3) Non existing drives: difference is in Sectors per Cluster: in FreeDOS: all output seems to be "undefined", in other OSes is all "undefined" except Sectors per Custer which seems to be always 255 (FFh) 4) Other diferencies are in CD/DVD roms and flashdisks. >From the above is shown that MS-DOS and WinXP give undoubtebly wrong results. The output for non existing drives seems to be random except AL, which was in all tested OSes except FreeDOS FFh. RBIL mentions it so maybe it could be fixed, however I think that programmers just shouldn't call this function for non existing disks and I even think this function shouldn't be called at all. -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
[Freedos-devel] bug in freedos kernel int 21.1c invalid drive handling?
Hi, I noticed that int 21 with ah=1c might return wrong values for "invalid drive", can somebody check this? For example: Is this a bug? When did we introduce it? What do other DOSes do here...? For invalid drives, I get: ds:bx=0:0 al=0 cx=400 dx=unchanged I expect: al=ff, other values are not explained in RBIL AL is sectors per cluster, CX is sector size, DX drive size in clusters or -1 if above 64k, DS:BX points to media ID byte Thanks for checking :-) Eric -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
