@ AL13N

Is it still like that, or is this solved in current cauldron?

Keywords: (none) => NEEDINFO
CC: (none) => marja11

no cauldron on that machine, but the mageia1 has this issue. (note that this issue may have selected cauldron, but it was before mageia1 release.)

@ AL13N

To be honest, I'm very ignorant and I do not know what this bug is about.
 
* When you speak about 2 devices, what do you mean: two video cards, two monitors, two something else?
* Why did you put a "?" behind nvidia in step one?
* What is "a mess" in step 4, please describe what happens

Version: Cauldron => 1

well, i have a laptop which has a sort of hybrid dual nvidia chipset.

but imho in general the option stands, if you have 2 cards, and choose nvidia over nouveau, you'll want ALL your nvidia cards to use nvidia module, because it would even conflict badly using both modules at the same time...

* so it's not 2 cards, but it sort of is 2 chips, in xorg.conf there are 2 "device" sections
* because i have a nvidia, but i would assume a similar problem would exist with any video card having multiple driver possibilities. (i want to show that i have nvidia, but i don't think nvidia would be required to test)
* the mess is what happens when nvidia and nouveau are loaded at the same time in the kernel (you don't want this to happen), you can test by checking if xorg.conf has 2 device sections and if they are using different drivers (but are still both the same brand)

in general, when you configure X, you configure them both at the same time. so i would expect both to be using the driver i chose, which is not the case.

@ AL13N

Thanks for clarifying. Is that your famous Acer laptop again?

Please attach the xorg.conf of when it went wrong.

indeed :-)

i don't have the results from back then anymore, but it's very easy to spot:

in the device section the driver starts by being nouveau (for both) and after doing XFdrake and choosing different driver, only one of both are being set to nvidia (i think it was the first one)

of course, to be able to test this, one would need multiple devices of the same brand.

I must say I nearly asked the Acer guy from bug 1695, but I can't find there are any Acer Aspire 2920's with a dual video card like yours. 

Just for the record, your Acer is a:

Acer Aspire 7530G Series - 724G32MN with a dual nvidia video card

And we are waiting for someone to reproduce this bug and for an xorg.conf of such a system with a dual video card, where only half the card switched to the new driver after choosing a different one.

Also not done yet: checked to see if this problem occurs in other linux distro's

Summary: XFdrake only modifies first devices when switching between drivers => Dual video card: XFdrake only modifies half the card when switching between drivers

tbh, I think you're missing the point here

this is about XFdrake (which is imho mageia only (except for mandriva, which had the same issue in the past)).

it's very easy, due to both devices being of the same brand, they need to have the same kernel module (or the different modules will conflict).

XFdrake modifies the xorg.conf when choosing different driver (priopriatary or not).

since XFdrake is handling both devices at the same time, it should modify both devices and not only 1 of them.

for an XFdrake developer this is likely a small issue, i could likely fix it myself with some pointers from XFdrake developers...

(In reply to comment #8)
> tbh, I think you're missing the point here
> 

Undoubtedly, I don't know anything at all, so I leave every possibility open, even including the issue being related to something being very special about that nvidia dual videocard

> this is about XFdrake (which is imho mageia only (except for mandriva, which
> had the same issue in the past)).
> 
If you mean the bug occurred there too: please give a link to that bug report

> it's very easy
> 
> for an XFdrake developer this is likely a small issue, i could likely fix it
> myself with some pointers from XFdrake developers...

Why don't you ask them?

(In reply to comment #9)
> (In reply to comment #8)
> > tbh, I think you're missing the point here
> > 
> 
> Undoubtedly, I don't know anything at all, so I leave every possibility open,
> even including the issue being related to something being very special about
> that nvidia dual videocard

i doubt it

> > this is about XFdrake (which is imho mageia only (except for mandriva, which
> > had the same issue in the past)).
> > 
> If you mean the bug occurred there too: please give a link to that bug report
> 
> > it's very easy
> > 
> > for an XFdrake developer this is likely a small issue, i could likely fix it
> > myself with some pointers from XFdrake developers...
> 
> Why don't you ask them?

I don't know who's the XFdrake developers, i would guess that when all info is supplied (it is now), that qa would assign to XFdrake maintainer?

The relevant developers are all CCd to this report.

Proper fix (handling of multiple graphics cards) is probably quite some work, and I'm not so sure this specific case could be easily solved with a hack. If the user selects a graphics card from the list, we simply do not know which card does he mean and which driver should be switched.

hmm, I had thought finding which videocard you have wouldn't take me a minute, but it took a lot more. Is should have asked you to provide the information.

Is NVIDIA® GeForce® 9300M  the correct one?

(In reply to comment #12)
> hmm, I had thought finding which videocard you have wouldn't take me a minute,
> but it took a lot more. Is should have asked you to provide the information.
> 
> Is NVIDIA® GeForce® 9300M  the correct one?

And now I remember, I once bought a system that contained different hardware than the specifications said.

I would really appreciate it if you'd provide the relevant part of the lspcidrake -v output

That is to complete this bug report as much as possible

it's actually marketed as "9400M SLI Hybrid", but i seems to contain both an "9300M GS" and a "9100M G"

lspci output:
Card:NVIDIA GeForce 6100 to GeForce 360: nVidia Corporation|G98 [GeForce 9300M GS] [DISPLAY_VGA] (vendor:10de device:06e9 subv:1025 subd:014d) (rev: a1)
Card:NVIDIA GeForce 6100 to GeForce 360: nVidia Corporation|C77 [GeForce 9100M G] [DISPLAY_VGA] (vendor:10de device:0844 subv:1025 subd:014d) (rev: a2)

(In reply to comment #11)
> Proper fix (handling of multiple graphics cards) is probably quite some work,
> and I'm not so sure this specific case could be easily solved with a hack. If
> the user selects a graphics card from the list, we simply do not know which
> card does he mean and which driver should be switched.

This seems indeed to be alot of work, but imho is this a feature request and not a bug.

I'm just asking that if you're going to replace a driver with another parsed in the configuration file, that you'll do this for all such drivers and not just once.

we can fix this bug without needing the feature request, imho, also see bug 200 , which is slightly related

(In reply to comment #16)
> (In reply to comment #11)
> > Proper fix (handling of multiple graphics cards) is probably quite some work,
> > and I'm not so sure this specific case could be easily solved with a hack. If
> > the user selects a graphics card from the list, we simply do not know which
> > card does he mean and which driver should be switched.
> 
> This seems indeed to be alot of work, but imho is this a feature request and
> not a bug.
> 
> I'm just asking that if you're going to replace a driver with another parsed in
> the configuration file, that you'll do this for all such drivers and not just
> once.
> 

Setting priority to "enhancement" and version to "cauldron" because this is a feature request.

Assigning to maintainer.

Version: 1 => Cauldron
Severity: major => enhancement
Keywords: NEEDINFO => (none)
Hardware: i586 => All
Assignee: bugsquad => thierry.vignaud

ok, rereading it, it seems i'm not clear.

i meant that what #11 proposes is a feature request enhancement. i filed this with the purpose of quick fixing the bug, because that's easier.

if just all the drivers matching "old" are modifed into "new" instead of just modifying only one of them, that's a good enough solution (for now)

really handling both and redesigning the whole thing, is imho alot of work and likely will not be implemented or in any case it'll take a long time to do it.

can we just not get this FIXED in mga2 please?

if you change driver, in the parsing code, the s/old/new/ thing just has to be s/old/new/g so that all the old ones are becoming new and drivers don't conflict.

There is no s/old/new/ thing in the parsing code so it is not so trivial.

However, I do agree this is an actual bug, not an enhancement, so changing severity.

Severity: enhancement => normal

confirmed on mga3 RC

Still valid in Mageia 5RC I suppose AL13N?

Keywords: (none) => NEEDINFO

(In reply to Samuel Verschelde from comment #24)
> Still valid in Mageia 5RC I suppose AL13N?

I assume it is still valid today, pre Mageia 8, because of:

 (In reply to Anssi Hannula from comment #11)
> The relevant developers are all CCd to this report.
> 
> Proper fix (handling of multiple graphics cards) is probably quite some
> work, and I'm not so sure this specific case could be easily solved with a
> hack. If the user selects a graphics card from the list, we simply do not
> know which card does he mean and which driver should be switched.

And because I can't find anything in the git log of drakx-kbd-mouse-x11 that gives the impression this issue was fixed.

So removing NEEDINFO and CC'ing the Mageia Tools maintainers

Keywords: NEEDINFO => (none)
CC: (none) => mageiatools