Bug 74974 (i845)
Summary: | i845 video hardware is not supported by XFree86 4.2.x | ||
---|---|---|---|
Product: | [Retired] Red Hat Linux | Reporter: | Leonid Kanter <leon> |
Component: | XFree86 | Assignee: | Mike A. Harris <mharris> |
Status: | CLOSED NOTABUG | QA Contact: | David Lawrence <dkl> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 8.0 | CC: | bruceyucc, cpeppler, j, macgyver, puffpd, shay.mcloughlin, tim+redhat.com, uday_iitm, yanfali |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | i386 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2002-12-14 15:39:52 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Leonid Kanter
2002-10-03 10:26:08 UTC
i845 is officially unsupported. The driver is provided only as an end user convenience in hopes that it works for some people rather than leave them with no video at all. I backported this driver from CVS, and it works for many, but not all people, and it does not affect the rest of the supported hardware, so I decided to leave it in, so that the people it worked for, would be able to use it at least. Intel i845 video hardware will be officially supported, both in 2D and in 3D when XFree86 4.3.0 comes out. Sorry I can't do anything about this for now. Closing WONTFIX. Since people are frequently reporting issues with i845 video, and it is unsupported as stated above, I've decided to make this bug report the master bug to close future duplicates against. *** Bug 77381 has been marked as a duplicate of this bug. *** *** Bug 79405 has been marked as a duplicate of this bug. *** Info update: A lot of bug reports, and emails come in asking about i845 video, so I am summarizing things here so there is a common place for answers to issues regarding i845 video support. Some common questions I am asked are: Q) When will Red Hat Linux 8.0 have i845 support? A) Red Hat Linux 8.0 will not likely ever support i845 video hardware. The first release of XFree86 to support this hardware will be XFree86 4.3.0 as mentioned above, which will appear in the next release of Red Hat Linux. There are no plans of ever releasing an XFree86 4.3.0 update for Red Hat Linux 8.0, and no plans of backporting the i845 support from 4.3.0 to 4.2.x. Q) Why doesn't Red Hat configure i845 cards to use "vesa" A) During Red Hat Linux 8.0 development, initially we defaulted the Intel i845 video hardware to using the "vesa" driver. This driver is dependant on the hardware's VESA VBE BIOS to work correctly. User feedback showed that these cards do not work well at all with the "vesa" driver, and that it would be unacceptable to default to using "vesa" even if it happens to work for some people. We can't fix broken video BIOSs and so we can't make "vesa" work on something it doesn't work properly on. Q) Why doesn't Red Hat configure i845 cards to use "fbdev" driver A) The fbdev driver requires that your kernel is running with a configured framebuffer as per the Framebuffer-HOWTO. We provide this support in our kernel for users to use if they prefer, and XFree86 has fbdev provided as well. It is possible for users to configure their system to use the XFree86 "fbdev" driver with the kernel framebuffer support by following the instructions in the Framebuffer-HOWTO, however while we provide this for user convenience, we do not officially support this. It may or may not work, and users are free to try it however. redhat-config-xfree86 can be supplied "fbdev" as a driver name to test this once the user has manually configured their kernel to use a framebuffer device as per the aforementioned HOWTO. There is no plan for the config tools to ever support configuration of modules.conf for setting up Framebuffer devices, so that step must be performed by the user manually. And again, all of this just might work - but is officially completely unsupported. Q) Someone else I know seems to have gotten i845 working in RHL 8.0 with the standard i810 driver. If this is unsupported, how is this possible? A) The reason this is possible, is that I personally and voluntarily backported code for i845 from an earlier state of 4.3.0 development codebase (before 4.2.99.1) in my own personal spare time. It is very minimal beginnings of i845 developmental code, and was found to work for some people. Since it worked for more people than the "vesa" driver did (vesa worked for nobody), and since it did not harm support for any other hardware supported by the driver, I left the patch in place in Red Hat Linux 8.0 in hopes it would work for someone out there. It does work for some people aparently, however wether it works or not, it is merely provided for convenience to those it happens to work for, and it is not supported by Red Hat, nor by myself. I do not own any Intel video hardware, and will not touch i845 in 4.2.0/4.2.1 any further since it is officially unsupported and advertised as such, and XFree86 4.3.0 will have support. It would be a huge waste of my development time (or my personal time) to spend on any such effort. Q) How can I get i845 to work "unofficially" somehow? A) In addition to the information supplied above, Intel i845 code is in XFree86 CVS right now, and that code is available in Red Hat Rawhide as developmental software in a constant state of developmental flux, both XFree86.org developmental, as well as Red Hat developmental. Users who are willing to personally accept the risks of using such unstable development code, are free to give it a shot, but should not do so on any important production systems or systems containing important data, as rawhide packages are not official releases, and are not quality tested. WARNING: DATA CORRUPTION IS POSSIBLE FROM USING THESE PACKAGES. At the time of this writing, i830/i845 support in rawhide and XFree86 CVS is broken, and requires both X side and kernel side fixes which will likely be coming before long. When that occurs is anyone's guess, I have no idea however. Q) Can I email you personally for help trying to install or configure rawhide XFree86, the kernel or anything else? A) Please do not do that as I do not have the time to help hundreds of individual people out with things like this, particularly when it is doomed to failure from the beginning due to the nature of developmental code. If you are brave enough to risk using the experimental packages, please subscribe to XFree86.org's mailing lists, dri-devel mailing list on sourceforge, and to Red Hat's own mailing lists, and ask for help there where hundreds of others are available that will try to help you. My time is best spent hacking on XFree86, so I'm likely to not respond to direct inquiries for assistance. I hope these Q&A's above answer the many questions people have been wondering about i845 support in XFree86. Again, please do not email me personally asking me questions about this. I will update this bug report with any new information I get as time goes on. Once I have reason to believe the driver is even semi-stable in rawhide, I will mention it here as well. XFree86 4.3.0 is slated for release in the end of January, so it is likely i845 will work by then. Hope this helps, Thanks. *** Bug 79494 has been marked as a duplicate of this bug. *** Thank you for all your hardwork in this area. Filing the bug wasn't meant as a criticism. Everyone benefits from your hardwork and dedication to making redhat a top flight distribution. I apologise if any bugs I filed made me sound ungrateful, I'm not. I have always known running Linux on 'brand new' hardware is a always a dodgy prospect and being one or two generations behind is always a safer move. Though it seems if you throw hardware at Alan Cox, he fixes things out of sheer bloody stubborn mindedness :> I think more hardware companys should just give RedHat early release hardware, just like they do with MS. Maybe one day. Thanks Again. I see that this is 'closed wontfix'. Can I suggest that there's at least some easy to find warning about this hardware on the RH site: it seems mostly possible to work around the problem or nail that you can't with particular hardware. Raising awareness of the problem with possible fixes dramatically improves the service to customers of the distro. in my experience the alternative is frustration and annoyance for the customer. Yan-Fa Li) No criticism was taken from your bug report. Just understand that Red Hat is not a video card company, nor a video driver company. We provide the video drivers that are available from the XFree86 project, and thus only the hardware supported by the XFree86 project is supported by Red Hat as well. Red Hat also contributes to the XFree86 project whatever bug fixes and enhancements that we develop. It is important to note however that it is not Red Hat's responsibility to develop video drivers for video hardware. In the case of the Intel i845, Intel and Tungsten Graphics have written the video driver support for this chipset. Since the current release of XFree86 in Red Hat Linux 8.0 is 4.2.0, and our 4.2.0 is identical to the since released 4.2.1 release driverwise, we are shipping support for every chip supported by an official XFree86 release currently. The existing i845 code that was implemented recently by Tungsten Graphics is not stable, and is not compatible with XFree86 4.2.x. It is impossible to release support for this chip until it is completly stable, and officially released by XFree86.org It would not at all make any difference if Intel were to ship Red Hat video hardware or not, as we are not a video driver shop that writes video drivers for free. It is the responsibility of video hardware manufacturers to either develop video drivers for their hardware, or to provide documentation to the community to do so, or to fund independant contractors to develop video drivers. It is not the responsibility of Red Hat to produce drivers for hardware, and even if we wanted to do so, and had the manpower and resources, the documentation for the Intel video hardware was not publically or privately available from Intel in any case, and I'm not even sure if it is available to us even at this date. It is Intel's responsibility to write video drivers for their video hardware for the platforms that they wish their hardware to be supported on, or to pay someone else to do so. Tim Coote) While I certainly appreciate your opinion, if hardware is not listed as supported, then it is not supported. How are we supposed to keep track of all the video hardware out there that isn't supported, and then specifically list somewhere "this isn't supported"? That is impossible. We do know what hardware XFree86 does support, because XFree86 publishes a status document listing the supported hardware. Before purchasing hardware, users should ensure that it is in fact supported by the operating system they are choosing. To all) I spent my own personal time trying to backport minimal support for this video adaptor into the Red Hat Linux 8.0 release in hopes that it would work for some users. It is negative comments like the last one, that make me want to not bother trying to do so in the future. It is much simpler for me to just state that a given card is not supported and leave it at that. So far in my experience, it is frustration and annoyance for the developer. In conclusion: You should research hardware support before you open your wallet, and unless you see explicit claims of support for something, assume it is not supported. I'm reopening this bug report due to the above mention of it being closed WONTFIX. I'm closing it now as NOTABUG, because in fact, it is not a bug at all that i845 is not supported. Support for the video adaptor would be a feature, and it is this feature which is not supported by XFree86 4.2.0. Lack of a feature, is not a bug. Reclosing as NOTABUG. Mike Of course you are strictly correct that one should check whether a piece of kit is supported before purchasing, that simple to cover everything that's out there and also that this issue is a feature of XFree86. However, - RH's value proposition is to manage the distro's for its customers to make their life easy - anything in the underlying products directly affects RH's brand. - Customers have been aware of this issue for some time, but you have to navigate thro' bugzilla to find it, and there's nothing that I can see to suggest that that's where a punter with a problem should have a quick look for known issues. - In practice, it's not easy to point check all hardware support ahead of purchase - the stuff that I bought was stock low end Dell kit with an Intel graphics adapter. Not exactly some dodgy fly-by-night pair of enterprises, nor very bleeding edge. - Personally, I feel that I'm pretty tolerant of this kind of feature in RH provided code. But when something doesn't work as expected I need to be able to eliminate dodgy hardware and finger trouble as soon as I can. I felt that simply making available the fact that other users had experienced difficulties with related configurations would be a big benefit to RH's customers, at little cost, and would add to the brand value. RH's choice, of course, and clearly you need to avoid scenarios I've seen with some distros where reported problems don't get back to the original package maintainers. tc Tim Coote) Unfortunately however, your electrons are wasted because I am an engineer. I make bugfixes to code that is buggy when I can do so. I try to support drivers that need work. I try to add features that are feasible and stable to add. I do not however work in marketing, tech support, or on the Red Hat website, and none of the people who do are reading your messages. So, while you can share your opinion here in bugzilla, it is really not getting anywhere that it would matter ultimately. I am doing my job, by telling people what is and is not supported, and I am trying to support what is supported. I also try to add support for some hardware that is not supported and make it supported - time permitting. So while I can appreciate your comments, they're not really hitting ears of those that make decisions. Your comments are best directed not at bugzilla, but at telephone lines leading into Red Hat, to make your voice known and counted. Hope this helps. Mike) I think that you make my point well ;-) is there any internet connected channel that you could recommend - I don't wish to hang on the 'phone, or pay the bill from the uk. tc Hello mharris, actually in the past RedHat has intervened in video driver problems. Neomagic comes to mind. Secondly, I believe that RedHat probably should get early hardware (and programming specs too for good measure), because that's how problems are spotted in the first place. The evil empire tends to get early hardware and this makes for a better experience for us users. I don't want to get into an argument with you over this as you are obviously time constrained and very stressed. RH8 works remarkably well and has an excellent out of the box experience. I'm sure it will get even better as time goes by, in no small thanks to people like you. However, your last point on not being the right person for all the comments here. Couldn't you just forward it to the right people ? Visibility might help your work load a tad and reduce some of your stress. Just a thought. Maybe RedHat needs a marketing bit bucket for us users to send suggestions into. *** Bug 80183 has been marked as a duplicate of this bug. *** Thanks for the all the info about the status of the video card support and for explaining that monitors are basically dumb and that they are not supported as such. It was exactly the help I needed. I got a supported Video Card and set the bios to use the second card as the default video card on bootup. 'redhat-config-xfree86' seemed to have problems when there were two video cards giving an error about on index out of bounds from the line self.resolution = d.modes[0].name in the script /usr/share/redhat-config-xfree86/xconf.py at about line 267. This line was in a 'for' loop which I changed to for d in screen.display: print d if d.depth == self.colordepth: print d.depth self.resolution = d.modes[0].name break so that the loop would only execute once The command I then used was redhat-config-xfree86 --set-card=s3virge --set-videoram=8192 -v and it worked fine and my new fancy 43B1-M monitor is now working. Thanks again, hope this additional info is of use to others i am able to get to the GUI now after running xf86config but now i have a problem with the mouse.... i have tried running mouseconfig ...but in vain the GUI is working perfectly but the mouse pointer just moves hay wire...when i move the mouse..... i am able to exploit all the features in the GUI using my keyboard... help me in this regard *** Bug 80183 has been marked as a duplicate of this bug. *** puffpd) As stated in bug #80183 - bugzilla is not a technical support forum or a place to get configuration and end user help. If you require help configuring XFree86 or anything else, join the xpert mailing list and ask your question there. Bugzilla is strictly for bug/defect tracking. Thanks. *** Bug 81405 has been marked as a duplicate of this bug. *** *** Bug 82145 has been marked as a duplicate of this bug. *** Adding bug alias "i845" to report to simplify closing dupes. *** Bug 82936 has been marked as a duplicate of this bug. *** *** Bug 85699 has been marked as a duplicate of this bug. *** |