DosFs 2.0 Product Information
Don Dewar
for it. When I pointed out I heard it was free with a support contract he
relented and sent me the binaries. They charge an outrageous amount of
money for the sources.
The integration went ok, but we suddenly found some of our files
disappearing or getting corrupt. We tried to debug it for months with very
little support from WRS. Finally, we got the sources and still could not
figure it out. When we compiled with less than 2 MB support all our
problems all vanished.
Anyway they don't seem to support it very well.
Good luck.
-- Don
"Leonid Rosenboim" <Dont_Spa...@Consultant.com> wrote in message
news:newscache$w61gog$hug$1...@lnews.actcom.co.il...
> I tried to find the technical specifications of DosFs 2.0 on the WindRiver
> Web site to show someone, and to my surprise, I found no mention of it.
> This seems shocking to me that this optional component is not advertised,
> and I remain puzzled as to the reasons why.
>
> Surely I don't expect any official answer from WindRiver (although that
> might
> be nice to chat again with my old pals), but I would appreciate if anyone
> can
> shed some light on this enigma.
>
> Of course this is mere curiosity.
> - Leonid
>
>
>
Leonid Rosenboim
"Don Dewar" <dde...@gothamnetworks.com> wrote in message
news:aonT7.93$5f4....@dfiatx1-snr1.gtei.net...
> [snip] Finally, we got the sources and still could not
> figure it out. When we compiled with less than 2 MB support all our
Would you care to elaborate on this ?
Maybe we can figure out the cause of trouble and help other poor souls who
may be stepping into the same sh*t ?
Charles Johnson
size of files would suddening become -1. We had to go to recompiling
the dosFs2.0 source at a lower optimization level to get it to work.
(WRS finally issued SPR #71089 for this.)
I would like to see WRS do a more fault resilient file system than
DOS.
Charles Johnson
Oresis Communications, Inc.
cjoh...@oresis.com
Andray Kaganovsky
@dfiatx1-snr1.gtei.net:
[AndrayK]. As far as I know, dosFs-2 is included with both Tornado-2.1
and Tornado-3.1, and does not exists as separate product anymore.
Andray
Leonid Rosenboim
ALl I found about file systems in the datasheet (for both Vxworks and AE) is
this:
Fast, flexible I/O and local file system
POSIX asynchronous I/O and directory handling
SCSI support
MS-DOS-compatible file system
TrueFFS flash file system (optional)
ISO 9660 CD-ROM file system
PCMCIA support
Not much to go on, and no mention about which dosFs is bundled. I heard that
AE includes DosFs 2.0, but again, no proof or detail in the literature.
Also, try to search for specific file system features in the Web site, there
is nothing ...
My only conclusion is that WRS marketing did not improve much in the last 3
years.
Leonid
"Andray Kaganovsky" <andrey...@home.com> wrote in message
[snip]
Werner Schiendl
"Leonid Rosenboim" <Dont_Spa...@Consultant.com> wrote in message
...
>
> My only conclusion is that WRS marketing did not improve much in the last
3
> years.
>
As - so it seems - has the file system.
- Werner
Leonid Rosenboim
but from your description of the work around, I might try to guess
that this has something to do with 64 bit math. DosFs 2.0, uses 64 bit
integer arithmetics, and does need the GCC and associated libraries to
perform these operations correctly. Trouble is that it is probably
the only module in VxWorks which uses 64 bit math, and therefore
this functionality is not properly tested after a product has been compiled,
and may fail on certain architectures.
If you finally got around by recompiling, this would suggest there is
nothing
wrong with the file system code itself, as the source code did not change
when
you recompiled.
Would appreciate more details, as I still care about this creation of mine.
- Leonid
"Charles Johnson" <cjoh...@oresis.com> wrote in message
news:fa9c3eaa.0112...@posting.google.com...
Leonid Rosenboim
"Werner Schiendl" <ws-...@gmx.at> wrote in message
news:3c20bca4$1...@brateggebdc5.br-automation.co.at...
Apparently, as the original team that developed dosFs2 has left WindRiver in
the beginning of 1999.
I have headed the team and resigned from WindRiver in January 1999.
But I do beleive that dosFs2 is an excellent product, many people are using
it without a problem, and I am still trying to figure out what went wrong in
those cases where people had problems.
- Leonid
Werner Schiendl
"Leonid Rosenboim" <Dont_Spa...@Consultant.com> wrote in message
news:newscache$172nog$af1$1...@lnews.actcom.co.il...
...
> But I do beleive that dosFs2 is an excellent product, many people are
using
> it without a problem, and I am still trying to figure out what went wrong
in
> those cases where people had problems.
>
From my point of view, the biggest problem with dosFs2 is just simply that
it is DOS filesystem.
There is absolutely no safety with this filesystem against all forms of data
loss.
Especially with read/write errors on the media or unexpected power loss.
All these problems are inherent to the way DOS (aka FAT) filesystem handle
file allocation, directories, etc.
This has nothing to do with the quality of the code, it is just a shame that
WRS has nothing better (= more stable and reliable) to offer here.
best regards
Werner
Leonid Rosenboim
First, I do not feel that the FAT disk structure is as bad as you describe
it. I have spent a good amount
of time thinking about the reliability issues, and I even recall writing a
white paper on this subject,
you can do good a solid stuff with this file system using rather simply
application design rules.
Many have done so succesfully, and are quite happily running mission
critical systems,
with MTBF far better then Windows and even Linux based systems of similar
role.
Secondly, one of the key goals for dosFs 2.0 was to REPLACE the older
dosFsLib,
which suffered many problems, and had key fratures missing, which Marketing
has identified
as important features to have - large disk drives with FAT32,
MSFT-compatible long names with VFAT,
built in consistency checker, smaller footprint for low-end designs, etc.
These improvements
where deemed much more important than completely new file systems.
There could have been more file systems in VxWorks, such as journaling file
systems,
but I felt that WInd River top level management did not regard file system
as a key
area where Wind River needs to excell, hence these good intentions did yield
any
products that one could use today. If at any time Wind River top guys change
their mind,
I'd love to be informed.
Regards,
- Leonid
"Werner Schiendl" <ws-...@gmx.at> wrote in message
news:3c21feb0$1...@brateggebdc5.br-automation.co.at...
Bob Waisnor
LONG time. Both Windriver and M-systems did the "Gee, it shouldn't do
that". Come to find out that the problem is in the dosfs 2.0 binary.
Here we re-compiled and did not use 2 Gig support and the thing works.
My guess it's a compile issue....
"Don Dewar" <dde...@gothamnetworks.com> wrote in message news:<aonT7.93$5f4....@dfiatx1-snr1.gtei.net>...
Leonid Rosenboim
+ What is your processor architecture ?
+ How did you disable 2Gig support before recompiling ?
"Bob Waisnor" <bwai...@gothamnetworks.com> wrote in message
news:52f21cc5.01122...@posting.google.com...
Leonid Rosenboim
DosFs 2.0 has a problem with Tornado 2.x on PowerPC 603/604 architecture
(or other derivatives using the 603/604 compule flag),
which is due to PowerPC compiler improper code generation for 64-bit
integer operations used throughout DosFs 2.0.
This bug could is tricky to catch, it will bite you only if you have
some tasks doing Floating Point calculation, and task switch taking
place between these and other tasks doing DosFs I/O in midst of
64-bit arithmetics. So do not be fooled to beleive you eint got it!
For those who got DosFs 2.0 source code, this can be solved by
either compiling the library with -msoft-float compile flag, or by
disabling 64-bit math (not recommended).
For binary only licensees, you need a patch which is available from
WINDSURF.
References:
SPR # 71089
Additional SPR # 33221,33702,33684,34704,62415,28428
"Bob Waisnor" <bwai...@gothamnetworks.com> wrote in message
news:52f21cc5.01122...@posting.google.com...