Community Packages

Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines

Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.

REPEAT: Do NOT report bugs for outdated packages!
Tasklist

FS#63476 - [beignet] broken on Bay Trail-T

Attached to Project: Community Packages
Opened by Alex Medvědů (satcom886) - Friday, 16 August 2019, 14:45 GMT
Last edited by Bruno Pagani (ArchangeGabriel) - Wednesday, 14 October 2020, 18:41 GMT
Task Type Bug Report
Category Packages
Status Closed
Assigned To Bruno Pagani (ArchangeGabriel)
Architecture x86_64
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 1
Private No

Details

Description:

Basically beignet is broken on the iGPU of a Lenovo G50-30. Each time I run anything in Hashcat I get a segfault. The CPU in the G50-30 is an Intel Celeron N2840 with Intel(R) HD Graphics Bay Trail-T (neofetch says Intel Atom Processor Z36xxx/Z37xxx Series Graphics & Display). The Github readme (https://github.com/intel/beignet#supported-targets) shows "3rd Generation Intel Atom Processors "BayTrail"" as a "supported target". I have another laptop with an I7-7700HQ and beignet works as it is supposed to. I am not sure this is something wrong with just the Arch build of the package or if the problem is in the upstream. I thought to report this here before opening a Github issue. One last note: the Ubuntu 18.04 build of the package worked flawlessly (not sure but since it is Ubuntu I suppose it was an earlier version).

Additional info:
* package version: 1.3.2+12+gfc5f430c-2
* I can provide any additional info if needed

Steps to reproduce:
1. Install beignet (from community) on the mentioned hardware
2. Try running any hashcat command (example hashcat -b)
3. After a few seconds you will end up with "[1] 3580 segmentation fault (core dumped) hashcat -b"
This task depends upon

Closed by  Bruno Pagani (ArchangeGabriel)
Wednesday, 14 October 2020, 18:41 GMT
Reason for closing:  Won't fix
Additional comments about closing:  beignet is being removed from the repository with the introduction of llvm 11. It will be added to the AUR, will free to fix it there by eventually building it against older clang/llvm.
Comment by Bruno Pagani (ArchangeGabriel) - Friday, 06 September 2019, 15:27 GMT
Beignet is dead and indeed full of bugs. For newer generation chips, there is Intel Compute Runtime, a.k.a. Neo. For older one, there is no hope.
Comment by Alex Medvědů (satcom886) - Sunday, 08 September 2019, 00:25 GMT
I know about the "Neo" runtime, but last time I checked it doesn't work with this chip.
Comment by Bruno Pagani (ArchangeGabriel) - Sunday, 08 September 2019, 08:50 GMT
Yes I know. Hence “newer” in my previous post. You are in the “no hope” category.
Comment by Alex Medvědů (satcom886) - Sunday, 08 September 2019, 09:09 GMT
Ok, thanks anyways...
Comment by M. Kaburagi (m.kaburagi) - Sunday, 08 March 2020, 14:08 GMT
Here are my findings:

1. beignet-git (from AUR) allows hashcat to work, probably because it is built against LLVM/Clang 7. Bug reporter (satcom886)'s comment on GitHub [1] also says "Hashcat works (at least the benchmark does until it gets to hashmode 1700 (SHA2-512))".

2. The aforementioned package, however, seems to cause hashcat to segfault in a certain case. In addition, darktable fails to start. (I can provide more details if needed.)

3. I referred to openSUSE's patches and build files for the beignet package [2] and rebuilt beignet, i.e. using the stable 1.3.2 release, applying the 4 patches, and using LLVM/Clang 7. Afterwards, hashcat and darktable problems mentioned above seem to be gone.

[1] https://github.com/intel/beignet/issues/15#issuecomment-562831284
[2] https://build.opensuse.org/package/show/science/beignet

Loading...