Logs
I have spent the day getting you some logs, I hope they are useful to you. I have split them into categories, each zip file contains a number of other zip files for the various categories. Logs include XORG, grep-radeon, grep-drm and dmesg.
1. Kernels without fixes using both cards no memory limitations
2. Kernels without fixes using both cards with memory limitations (mem=3500M)
3. Kernels with fixes using both cards no memory limitations
4. Kernels with fixes using both cards with memory limitations (mem=3500M)
5. Kernels with fixes using single card FirePro 2600 in 16 x PCIE slot no memory limitations
6. Kernels with fixes using single card FirePro 2600 in 16 x PCIE slot with memory limitations (mem=3500M)
Notes
Are the 2 images included in Alpha4 the same as both booted to my FirePro.
I could not get Alpha5-No-Radeon to boot under any circumstance, it hungs after reading kernel and tree on each occasion.
Using the single card every fixed kernel booted successfully to the FirePro in the 16 x PCIE slot
All unfixed (ie included latest DRM updates) booted successfully to my Southern Island Cards only.
Thanks for all your help and apologies for the multiple zipped files!
On 2019-01-01 4:03 p.m., Christian Zigotzky wrote:
Hello,
An end user reportet that his ATI FirePro 2260 doesn't work anymore with the latest Git kernel. We are a first level Linux support and we usually compile Git kernels during the merge window time for testing. After the update 'drm-next-2018-12-14', his ATI FirePro 2260 doesn't work anymore. We removed the update 'drm-next-2018-12-14' and after that he reported that his ATI FirePro 2260 works again.
I really wouldn't have a clue what I am looking for but... with Alpha5 (no fix), it appears that both cards are found and initiated but I read something in the XORG log that indicated one of the displays was not dri2 compatible not sure if that means anything. I'll have to test firepro only against non fixed alphas.
Christian
I have been thinking about this. Perhaps we are looking for an error where there may not be one, perhaps everything is working as it should (by today's standards). I checked an old thread regarding the use of dual cards and the trick was that by renaming libglamoregl and removing Xorg.conf the SI would not initiate and hence the second card would be used. Checking the Xorg logs I notice that the missing libglamoregl is logged but everything continues as normal but the SI card is found, probed for capabilities and initiated. So something has changed which removes the dependency on libglamoregl to boot to a SI card, not necessarily an error but perhaps a change in process for initiating GPUs.
Michel Dänzer wants that you file a bug report at bugs.freedesktop.org. You can wrote your thoughts there. The strange thing is, that your Firepro doesn’t work alone (without another graphics card) with the alphas in your X5000.