Re: [Freedos-user] UIDE versus UHDD in freedos 1.2 - and 1.3
Hi Rugxulo, Skipping over the FUD undertone, here are some answers :-) Yes the UHDD updates in 2019 are useful and by the author. Yes UHDD + UDVD2 works better than using only UIDE alone. No this is not about RDISK or XMGR updates, although if you can be more specific, I can check whether ibiblio needs them. Yes UHDD and UIDE have different performance features and UHDD is significantly better in various aspects. Yes UDVD2 is fine in spite of having no recent updates. Yes EMM386 can have troubles with new hardware. But: Yes Japheth is updating his EMM386 JEMM right now. You ask whether anybody should still use UIDE alone: Probably no. It is a bit smaller. Maybe floppy users. Yes the 2019 releases of the drivers are free and open. Regards, Eric ___ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
Re: [Freedos-user] UIDE versus UHDD in freedos 1.2 - and 1.3
Gruess Gott, On Mon, Mar 9, 2020 at 5:59 AM Eric Auer wrote: > > > As far as I'm concerned, UIDE [sic] died in 2015. > > Another reason to switch to UHDD and UDVD2. I meant that I'm only personally aware of the 2015 "drivers" (long ago abandoned). I was not directly informed of the later "ad hoc" early 2019 update of one, lonely file (UHDD.SYS). So I don't know the difference because I only ever ran UIDE.SYS (which apparently is defective or at least designed differently, according to you). Who even takes credit for the 2019 update? Certainly it can't be the original developer, can it? He still maintains his closed source variant (last updated in November 2019), for zero practical advantage. Why have so many competing variations (market segmentation??)? > You do know that the 2019 version *does* include full sources > of UDVD2, UHDD, UIDE and XMGR, I hope. AFAIK, the developer is aware of and has elsewhere fixed known bugs in RDISK and XMGR. But none of those fixes were propagated back to the 2015 version (UIDE, aka not XIDE) for FreeDOS on iBiblio. Why is that? > Replying to my list, you ask for more exact descriptions of the > improvements in the currently-on-ibiblio 2019 UHDD and UDVD2: > > Better performance: UHDD 10% faster with read-ahead than UIDE. Were these two designed differently? Or is this only in hindsight, i.e. some 2019-era bugfix of older 2015 code? Why have two separate versions that behave differently? (There could be a good reason, I just don't know why exactly.) > Because UHDD (and UDVD2, in spite of being "old") "Old" is fine when it works. I'm not complaining about age but moreso bugs, regressions, restrictions, licensing, redistribution. Also, having too many subvariants and releases is confusing. It shouldn't be so fragmented. > recognize more drives as DMA/UDMA compatible, without false > positives, they give much better performance in those cases > compared to situations where UIDE fails to detect the DMA support. > This can mean up to several *times* faster in EMM386 context, as > a BIOS would rarely bother to call the VDMA API to support > fast protected mode or VM86 disks on DMA and rather use PIO. I was under the impression that most EMM386s were unreliable on newer hardware. So I don't try to use them too much. (Then again, CSM is basically dead, so who cares.) Yes, Japheth has updated JEMM386 a few times in recent months. Is that the one you're referring to, or do you refer to other (much older, more limited, buggier) vendors? Which ones have been tested in recent years with these drivers? But even with JEMM we have several versions (5.78, 5.79, 5.80-pre1) and subvariants (e.g. jemmex or jemm386). I'm not complaining, we're lucky to have it, it's just a lot for people to test (but most people don't). So I haven't tried it lately. > > How is that even possible? Too many versions, too many > > (alleged) bug fixes! Ridiculous! > > Being too annoyed to look at the new version will not make > the new version worse. That is just your personal opinion. Is UIDE.SYS completely irrelevant in lieu of UHDD.SYS + UDVD2.SYS? Seriously, do you know? Is there literally any reason anymore to use UIDE.SYS? Were they just designed differently, or is this only a result of UHDD.SYS being "fixed" in early 2019?? > PS: The drivers are deliberately freeware with sources without > giving a specific license as the author is against fine print. FreeDOS, by design, has always favored free/libre or "open source". * http://wiki.freedos.org/wiki/index.php/Open_source_software * https://sourceforge.net/p/forge/documentation/userfaq/#free * https://www.gnu.org/philosophy/free-sw.en.html Going closed source for unjust reasons for zero practical advantage is user hostile and the exact kind of thing that the GPL was designed to avoid. No amount of partial updates and pretend lip service can change that. He made a choice, based upon nothing, despite correction and proof, but instead still keeps punishing end users, for literally no gain. What a waste of time. ___ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
Re: [Freedos-user] DMA error trying to format floppy...
Hi! If you get floppy DMA errors with FreeDOS FORMAT (which version? error message?) then consider avoiding UMBs (EMM386, UMBPCI) or using helpers such as LOWDMA of DOSLFN or DMACHECK of UMBPCI to ensure usability. There should also be separate helpers to protect floppy access from crossing 64k DMA boundaries, loadable as a TSR, but I have no specific name or URL around right now. Maybe somebody on the list has a link for you :-) > Trying to use the Freedos 1.3 live cd to get a dos prompt > and test formatting floppies. My hardware is special. > > I'm using a PICMG 1.0 passive backplane or more > specifically the HPP 14S backplane. > > I'm using an EVOC FSC-1740VNA single board computer. > > This SBC does not have an onboard floppy controller. > > I'm using an ISA multi I/O card with a floppy controller, > but note that the EVOC is designed for a USB floppy. ... > I'm getting some sort of DMA overrun trying to format a floppy in freedos... > > -- Michael Robinson ___ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
Re: [Freedos-user] UIDE versus UHDD in freedos 1.2 - and 1.3
Hi Jim, > I think this is an oversight in FreeDOS 1.3. I did review UDVD2 and the > others and said they were ok for FreeDOS 1.3. They are listed as such > (green) on the Packages page. > http://wiki.freedos.org/wiki/index.php/Releases/1.3/Packages > > Unless I'm missing something? I agree: It seems that while UDVD2 is included, there could be a more explicit mention and inclusion of UHDD and the fact that UDVD2 combined with UHDD is strongly recommended over using the integrated but older UIDE driver instead. So apparently we should simply add UHDD to FreeDOS 1.3 as UDVD2 already is there :-) Eric ___ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
Re: [Freedos-user] UIDE versus UHDD in freedos 1.2 - and 1.3
I think this is an oversight in FreeDOS 1.3. I did review UDVD2 and the others and said they were ok for FreeDOS 1.3. They are listed as such (green) on the Packages page. http://wiki.freedos.org/wiki/index.php/Releases/1.3/Packages Unless I'm missing something? Jim On Mon, Mar 9, 2020, 5:59 AM Eric Auer wrote: > > Hi Rugxulo, Jerome, Jim et al, > > > The full 1.2 release was from late 2016 / early 2017. > > It hasn't changed. > > Good point, although 1.3 is still pending, so people > might still want to update while they only have 1.2. > > > As far as I'm concerned, UIDE [sic] died in 2015. > > Another reason to switch to UHDD and UDVD2. > > > That makes no sense (to me). UHDD.SYS (from 2015) indeed had a > > surprise update in early 2019 (dunno what changed, ask Jim), but > > UDVD2.SYS is still dated 2015. > > > > * > https://www.ibiblio.org/pub/micro/pc-stuff/freedos/files/dos/cdrom/uide/ > > > > "It would probably be better" ... if XIDE/XHDD/XDVD2 (or whatever he > > calls it nowadays) wasn't unjustly "closed source" for five years! > > You do know that the 2019 version *does* include full sources > of UDVD2, UHDD, UIDE and XMGR, I hope. And I do *not* remember > attempts by other programmers to update them which would have > failed during the period when sources were unpublished, simply > because low level drivers rarely get updates from new people. > > Replying to my list, you ask for more exact descriptions of the > improvements in the currently-on-ibiblio 2019 UHDD and UDVD2: > > Better performance: UHDD 10% faster with read-ahead than UIDE. > > 386 compatibility: UHDD can run on 386, while UIDE follows old > Microsoft advice which causes XMS move errors on older 386 CPU. > > Improved drive detection and LBA: UHDD supports DMA on SSD (as > well as CF) which claim to be "ATA / ATAPI" while UIDE would > have ignored them as potentially optical, expecting *only* ATA > to be supported. UIDE supports only old LBA for the first 128 > GB, while UHDD supports LBA48 and larger disks. Note that DOS > itself has a 2 TB limit until somebody adds GPT partition code. > > Because UHDD (and UDVD2, in spite of being "old") recognize > more drives as DMA/UDMA compatible, without false positives, > they give much better performance in those cases compared to > situations where UIDE fails to detect the DMA support. This > can mean up to several *times* faster in EMM386 context, as > a BIOS would rarely bother to call the VDMA API to support > fast protected mode or VM86 disks on DMA and rather use PIO. > > > How is that even possible? Too many versions, too many > > (alleged) bug fixes! Ridiculous! > > Being too annoyed to look at the new version will not make > the new version worse. That is just your personal opinion. > > > These decisions (for FD 1.3) rely mostly on Jerome and Jim. > > Then I recommend UHDD and UDVD2 to Jerome and Jim, specifically. > > Regards, Eric > > PS: The drivers are deliberately freeware with sources without > giving a specific license as the author is against fine print. > > > > ___ > Freedos-user mailing list > Freedos-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/freedos-user > ___ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
Re: [Freedos-user] UIDE versus UHDD in freedos 1.2 - and 1.3
Hi Rugxulo, Jerome, Jim et al, > The full 1.2 release was from late 2016 / early 2017. > It hasn't changed. Good point, although 1.3 is still pending, so people might still want to update while they only have 1.2. > As far as I'm concerned, UIDE [sic] died in 2015. Another reason to switch to UHDD and UDVD2. > That makes no sense (to me). UHDD.SYS (from 2015) indeed had a > surprise update in early 2019 (dunno what changed, ask Jim), but > UDVD2.SYS is still dated 2015. > > * https://www.ibiblio.org/pub/micro/pc-stuff/freedos/files/dos/cdrom/uide/ > > "It would probably be better" ... if XIDE/XHDD/XDVD2 (or whatever he > calls it nowadays) wasn't unjustly "closed source" for five years! You do know that the 2019 version *does* include full sources of UDVD2, UHDD, UIDE and XMGR, I hope. And I do *not* remember attempts by other programmers to update them which would have failed during the period when sources were unpublished, simply because low level drivers rarely get updates from new people. Replying to my list, you ask for more exact descriptions of the improvements in the currently-on-ibiblio 2019 UHDD and UDVD2: Better performance: UHDD 10% faster with read-ahead than UIDE. 386 compatibility: UHDD can run on 386, while UIDE follows old Microsoft advice which causes XMS move errors on older 386 CPU. Improved drive detection and LBA: UHDD supports DMA on SSD (as well as CF) which claim to be "ATA / ATAPI" while UIDE would have ignored them as potentially optical, expecting *only* ATA to be supported. UIDE supports only old LBA for the first 128 GB, while UHDD supports LBA48 and larger disks. Note that DOS itself has a 2 TB limit until somebody adds GPT partition code. Because UHDD (and UDVD2, in spite of being "old") recognize more drives as DMA/UDMA compatible, without false positives, they give much better performance in those cases compared to situations where UIDE fails to detect the DMA support. This can mean up to several *times* faster in EMM386 context, as a BIOS would rarely bother to call the VDMA API to support fast protected mode or VM86 disks on DMA and rather use PIO. > How is that even possible? Too many versions, too many > (alleged) bug fixes! Ridiculous! Being too annoyed to look at the new version will not make the new version worse. That is just your personal opinion. > These decisions (for FD 1.3) rely mostly on Jerome and Jim. Then I recommend UHDD and UDVD2 to Jerome and Jim, specifically. Regards, Eric PS: The drivers are deliberately freeware with sources without giving a specific license as the author is against fine print. ___ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user