>>>>>> Great. Now take 1.88Mb floppy (maked with xdfcopy) and explain
>>>>>> how to create standard 1.44Mb floppy with mkfs.ext2 :-))
>>>>> mke2fs /dev/fd0H1440
>>>> Yes, really ?
>>> Yep...
>> Try before claim such things. 1.88Mb format is NOT compatible with
>> any standard format on low level (there are 1K sectors to start
>> with :-).
RW> In that case, what 1.88M format are you using? The one I use uses 512
RW> byte sectors, the same as are used by all the other floppy formats
RW> that I use.
There are "standard" 1.88Mb in OS/2. mtools has support for it. Really weird
format really: first track use 512 byte sectors, most tracks use 1K sectors, etc.
Looks like 2K floppy from Windows9x :-)
RW> If we're talking about different things, then naturally out results
RW> will differ...
:-))
>> So mk2efs will hanghs exactly the same way as in my experiment.
>> fdformat will work of course.
RW> Since your experiment turned out to be meaningless, so were its
RW> results...
>>>>>> Or even simpler: take 1.44Mb floppy and make 720Kb floppy from
>>>>>> it (this is safe).
>>>>> mke2fs /dev/fd0D720
>>>> Are you sure ? What about 5" 720K disks ? A have few 5" 1440Kb
>>>> disks and 720Kb disks :-))
>>> 5" 720k disks can be had with /dev/fd?h720 without problem. 5"
>>> 1440k disks are not one of the standard formats supported by
>>> Linux...
>> I have 5" 1440k disks and I'm could mount them as /dev/fd1H1440
>> just fine. mke2fs works with them just fine if I use /dev/fd1H1440
>> and does not work when I refer them as /dev/fd1h1200 !!! BTW 5"
>> 720k disks could be ACCESSED with /dev/fd?h720 but NOT CREATED from
>> 360k or 1.2M disks not via fdformat /dev/fd?h720 nor with mke2fs
>> /dev/fd?h720 !!! Both commands will just hungs !
RW> Curious that, wonder what you're doing wrong...
>>> Incidentally, my understanding as to why these work when using
>>> /dev/fd0 doesnae is because the suffixed versions over-ride the
>>> drive's own details and return their own...
>> Yes, suffixed version just says: "no, autodect will fail for this
>> drive -- do not try it". Suffixed version could not magically
>> reformat drive (fdformat could and here suffixed versions are
>> helpful since autodetect will fail here :-) but could just say to
>> kernel: "no there are not 18 sectors for track but only 9". Kernel
>> will happily use only 9 first sectors then. Floppy will look like
>> 720K floppy afterwards but still there are will be 18 sectors for
>> each track (9 of them not used). This is not a case for 1.88Mb
>> drive vs 1.44Mb drive (different sector sizes) nor for 1.2Mb drive
>> vs 360K drive (different number of tracks).
RW> Ah...that makes sense...
>>>>> I regularly do the latter, and have never had any problems with
>>>>> it...
>>>>>> Again format under DOS will do this just fine.
>>>>> Again, why bother with DOS ???
>>>> Since in DOS low-level format and filesystem creation is combined
>>>> in one program unlike Linux.
>>> True, but irrelevant...
>> VERY RELEVANT -- see below. DOS format could change format of media
>> if needed. mke2fs COULD NOT ! NEVER ! Interpretation of format for
>> already formatted media could be changed (sometimes successfull,
>> sometimes not) but reformating will NEVER occurs.
RW> I've now triple-checked, and I found some curious results...
RW> 1. 1.44M disk formatted as 720k with "mkdosfs /dev/fd0D720" is, as
RW> you state, actually formatted with 18 sectors per track, but
RW> only 9 of them used. However, Win95 has no problem using it as
RW> a 720k disk in that format.
RW> 2. 1.2M disk formatted as 720k with "mkdosfs /dev/fd0h720" is also
RW> formatted as 15 sectors per track with only 9 of them used, but
RW> in this case, Win95 is NOT able to use the resulting disk.
Confirmed. This is really weird since "real" 720k format (created with
pu_1700.com and MS DOS (any version including 7.x from Windows 9x (NOT from
GUI! Use F8 and "Command Prompt Only"): just run `pu_1700 /sm=2` from
ftp://ftp.sch57.msk.ru/users/khim/pu_1700/pu_1700.arj and then
`format b:/t:80/n:9` for 720K and `format b:/t:80/n:18` for 1440K) is ok both
with Windows 9x and Linux ! Even 830K (format b:/t:83/n:10) is ok with both
Linux (mtools, not just mount, unfortunatelly) and Windows 9x !!! I'm still
have quite a few 720K, 830K and 1440K floppies and use them extensively with
Windows9x (720K and 830K are DD floppies and 1440K are HD floppies;
1440 vs 1200 is not big difference but this is usefull to make diskcopy
(from MS DOS 7.x, not Windows9x) from 3" disks to 5" disks).
RW> 3. 1.2M disk formatted as 360k with "mkdosfs /dev/fd0d360" fails
RW> to format.
Understandable.
RW> I therefore stand corrected regarding my belief that mke2fs and
RW> mkdosfs were reformatting the disk completely, having been fooled by
RW> case (1) above, and I thus apologise for the resultant misleading
RW> comments...
RW> However, I do NOT apologise for the rest of my comments, which were
RW> apparently due to the two of us talking about different things...
>>>>> For reference, from this system:
>>>> Q>> $ ls -lFG /dev/fd0*
>>>> Q>> brw-rw-r-- 1 root 2, 0 May 5 1998 /dev/fd0
>>>> Q>> brw-rw-r-- 1 root 2, 12 May 5 1998 /dev/fd0D360
>>>> Q>> brw-rw-r-- 1 root 2, 16 May 5 1998 /dev/fd0D720
>>>> Q>> brw-rw-r-- 1 root 2, 28 May 5 1998 /dev/fd0H1440
>>>> Q>> brw-rw-r-- 1 root 2, 12 May 5 1998 /dev/fd0H360
>>>> Q>> brw-rw-r-- 1 root 2, 16 May 5 1998 /dev/fd0H720
>>>> Q>> brw-rw-r-- 1 root 2, 4 May 5 1998 /dev/fd0d360
>>>> Q>> brw-rw-r-- 1 root 2, 8 May 5 1998 /dev/fd0h1200
>>>> Q>> brw-rw-r-- 1 root 2, 20 May 5 1998 /dev/fd0h360
>>>> Q>> brw-rw-r-- 1 root 2, 24 May 5 1998 /dev/fd0h720
>>>> Q>> $
>>>> Ok. Just tried to use your braindead "theory" :
>> RW> ===8<=== Lots more rubbish omitted, relevant lines retained ===>8===
>>>> [root@localhost /root]# fdformat /dev/fd1h1200
>>>> [root@localhost /root]# mke2fs /dev/fd1h360
>>>> RedHat 5.1, /dev/fd1 -- 5" floppy drive, more information available
>>>> by request.
>>>> Why mke2fs works just fine after fdformat /dev/fd1h360 but not
>>>> after fdformat /dev/fd1h1200 ?
>>> Probably because those two are NOT compatible with each other.
>> Yes. It was EXACTLY what I'm want: two DIFFERENT, INCOMPATIBLE
>> formats but for the same physical media ! This is needed as
>> countersample for your stupid theory about magic format while
>> mke2fs ...
RW> Isn't it a pity the two you chose are NOT for the same physical media.
RW> The former is defined SPECIFICALLY for 3.5" floppies, the latter for
RW> 5" floppies, and whilst they behave identically on many systems, ON
RW> SOME THEY DO NOT!!!
RW> I have known for some considerable time that on two of my systems,
RW> /dev/fd0h360 and /dev/fd1d360 work fine, but /dev/fd0d360 and
RW> /dev/fd1h360 both fail, and on all the rest, all four will work quite
RW> happily.
This is VERY strange since both fd?d360 and fd?h360 are for 5" drives (double
and high density); fd?D360 and fd?H360 are for 3" drives (double and high
density -- in fact they are equal on hardware level :-)
>>> Try using the compatible options, which have been working fine for
>>> me for OVER TWO YEARS...and if they're not working for you, then
>>> either you've patched the kernel to prevent them from doing so,
>>> somebody else has done so, or the relevant entries in /dev have
>>> the wrong mknod numbers attached to them...
>> Yes, I know. But this will not prove ANYTHING about initial claim
>> !!! If there are compatible options then reformating is not needed !
RW> If they are HARDWARE-INCOMPATIBLE options like you chose, then you
RW> prove absolutely nothing, as you succeeded in doing above...
They are HARDWARE-COMPATIBLE (at least they are SHOULD BE hardware-compatible).
At least on my system they are really hardware compatible for sure...
>>> For reference, your second line should refer to
>> Q>> /dev/fd1d360
>> Q>> ^
>>> As I understand it, /dev/fd1h360 refers to a drive formatted as
>>> SINGLE sided, 80 tracks, 9 sectors per track, but I have never
>>> found a drive where it could be used...
>> You are joking. Yet again:
>> -- cut --
>> [root@localhost /root]# fdformat /dev/fd1h360
>> Double-sided, 40 tracks, 9 sec/track. Total capacity 360 kB.
>> Formatting ... done
>> Verifying ... done
>> -- cut --
RW> I have to admit that was a bit of a wind-up, to see whether you were
RW> as easily midled as you were pretending, and I'm glad to report that
RW> you're not...
:-))
RW> Incidentally, MS-DOS 3.10 was quite happy to use both 3.5" and 5.25"
RW> disks formatted to the specifications I gave above, and the MS-DOS
RW> format command could even be told to create them, but Linux has never
RW> (as far as I know) included support for them...
RW> For some reason, MS-DOS 5.00 and later's format command refuses to use
RW> that format...
I'm does not have MS-DOS 3.10 here but I'll try to check with 3.30 :-))
Which parameters for format there are should be ?
>> What's the difference between /dev/fd?h360 and /dev/fd?d360 ??? You
>> are joking ! This is just mean: high-density floppy formatted for
>> 360k and double-density drive formatted for 360k !!! (Why the
>> difference ? Oh. Long story. There are difference in coercive force
>> for different types of floppies :-)
RW> There's more of a difference than that, as my experience has shown.
RW> No, I don't know what the other differences are, but they clearly
RW> exist...
>> Just standard 2.0.34 kernel from RedHat 5.1 (I'm use some 2.1.x
>> kernel usually but specially for testing I'm reboted with this old
>> one :-)
RW> No problem. I'm using 2.0.36 myself, having upgraded RedHat 5.1 a
RW> while back...
>> Here was you point: mk2efs could format drive. This is HORRIBLE
>> wrong.
RW> As stated above, I accept that I was wrong about this, having been
RW> misled by some curious anomolies in the way Linux does things...
>> I am have physical media (5" disk :-) which could be formatted by
>> two different incompatible low-level format versions: as h1200 and
>> as h360 (d360 will work as well -- just information will be kept
>> safely on disk for less time :-).
RW> As stated above, fd?d360 and fd?h360 are NOT always identical, hence
RW> my querying your use of (to me) hardware incompatible formats...
Yes, they are not compatible but both fd?d360 and fd?h360 are for 5" drives !
fd?d360 is for double-density floppies and fd?h360 are for high-density ...
fd?D360 and fd?H360 are for 3" drives...
>> If mk2efs could format media (this was your initial claim and start
>> of discussion) then this should not lead to any problems.
RW> As stated above, I was wrong here, and I apologise for it...
>> This is not a case: if disk is formatted as h360 (via fdformat)
>> mke2fs works, if disk is formatted as h1200 (via fdformat) then
>> mke2fs hangs horrible. What this means ? Just one thing: mke2fs
>> will not format disk ! NEVER !
RW> Fair enough, although you may find my comments earlier in this missive
RW> of interest...
Hm. I'm still could not understood why "real" 720K (or 830K :-) drive is ok
with both Windows9x and Linux while simulated one is ok with Linux but not with
Windows 9x ...
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/