How ironical it is.. When I bought my first card(I have two of those broken
cards), GTX750(which AW has) didn't exist yet.
When I bought the second - GTX750Ti was out and i thought "naah, I'll go
crossfire route".
Half a year later I've stumbled on this very long archlinux thread with
success stories in hundreds of posts, and I thought it would be awesome to
try this stuff on my hardware. You know the results. It borked and forked
but ONCE worked. Very glitchy, but worked via VGA and seabios. To solve
those glitches i tried OVMF, you know the result.
Well, I kinda considered selling those cards and getting some other, but,
you know, when you begin you can't stop digging.
Seems like I'll continue digging again, maybe I'll find something. Thanks
for clearing this mess out, now I know that I am definitely damned by those
cards and ASUS.
Oh, maybe I could poke ASUS support again, but the cards are EOL now..
2015-02-19 20:36 GMT+03:00 Laszlo Ersek <[email protected]>:
> On 02/19/15 18:21, Blank Field wrote:
> >
> >
> > 2015-02-19 20:02 GMT+03:00 Laszlo Ersek <[email protected]
> > <mailto:[email protected]>>:
> >
> > (If you dropped the list on purpose, please don't do that; but maybe
> > your message is just withheld for moderation, if you are not
> subscribed
> > and/or attached a file.) So,
> >
> > On 02/19/15 17:32, Blank Field wrote:
> > >
> > >
> > > 2015-02-19 19:20 GMT+03:00 Laszlo Ersek <[email protected]
> > <mailto:[email protected]>
> > > <mailto:[email protected] <mailto:[email protected]>>>:
> >
> > > Indeed, an assert has failed exactly after finding PCI VGA device.
> > > Here's the full log:
> > > http://pastebin.com/vkKTu5Aq
> > > ASSERT
> > >
> >
>
> /var/lib/jenkins/jobs/edk2/workspace/rpmbuild/rpm/BUILD/edk2.git/MdeModulePkg/Core/Dxe/Mem/Pool.c(475):
> > > CR has Bad Signature
> > > What does this means?..
> >
> > Bad things.
> >
> > "Found PCI VGA device" (from the log) is okay, it comes from
> > "OvmfPkg/Library/PlatformBdsLib/BdsPlatform.c".
> >
> > The assert is very bad OTOH. It means that something has corrupted
> the
> > internal structures of the edk2 DXE memory allocator. In edk2 many
> > structures have embedded magic numbers (signatures) and whenever a
> > "narrowing" cast (a downcast) is done, like going from a contained
> > structure to a container structure, signatures are very frequently
> > checked.
> >
> > My take is that the UEFI oprom of your video card is borked. I'm
> basing
> > this simply on the fact that for many cards GPU passthru with OVMF
> works
> > (it works for me too :)). Does anyone else on the archlinux forum
> has a
> > similar card to yours? (Note that just because it doesn't cause
> > catastrophic failure on a physical UEFI platform, the oprom still may
> > very well be broken.)
> >
> > In any case, the next step would be to narrow down the bug.
> Instrument
> > the edk2 code with DEBUG statements, and see how far it gets. (Eg. it
> > could blow up when the card's UEFI oprom is run.) One file to
> instrument
> > could be "MdeModulePkg/Bus/Pci/PciBusDxe/PciOptionRomSupport.c", and
> > whatever calls its functions.
> >
> > Alternatively, using the gdb debugging tips I linked earlier, you
> could
> > grab a backtrace while in the dead loop.
> >
> > > Can i attach files to mails here?
> >
> > If it's plaintext, it shouldn't be a problem I think.
> >
> > Laszlo
> >
> > Seems like no one else on archlinux forum have those GPUs.
> > On the physical UEFI platform... system loads, but the image output is
> > frozen when the OS loads, but it continues to boot.
> > Chances of broken UEFI oprom are very high, because i've got that
> > firmware from asus' update tool a while ago, but i can no longer get it
> > there. Moreover, asus support tell me that "there were no bios recalls".
> > When i asked them for the freshest rom, oprom was not UEFI-compatible.
> > I remember i've found some unofficial UEFI-compatible oproms somewhere
> > on the net, but i can't find them anymore.
> >
> > I'll try building OVMF manually with debugging features, following the
> > guide by bcran you've provided, but this will be some time later.
> >
> > ..well, i guess i'm just very unlucky, since my vfio experiments always
> > blow up severely and very low-level.
>
> I believe in this case it originates from the card (and its oprom
> situation). When I decided to set this up for myself, the first question
> I asked (Alex, of course) was, what *new* card he recommended to buy.
> I'm usually extremely frugal (I tend to use hardware until material
> failure), but this stuff is very unconventional, so let's not maneuver
> ourselves into a corner immediately by starting out with the wrong card.
> The software issues are complex enough even when you have a known good
> card. I'm sure the archlinux forum has a number of recommendations.
>
> I'm sorry I couldn't help more.
>
> Thanks
> Laszlo
>
> >
> > The file attached is the troublesome ROM itself, it's binary and 128k.
> >
> > P.S.
> > damn that gmail interface, it's so counter-intuitive
>
>
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________
edk2-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/edk2-devel
