Re: [Debian-Stretch on macmini-4.1] Can't get X working with nvidia

[Ennio-Sr]

* Henning Follmann  [120318, 08:09]:
> On Sat, Mar 10, 2018 at 12:21:17AM +0100, Ennio-Sr wrote:
> > Hi all!
> > [...]

> > ...solution to the 'black screen' coming out running startx, I'm curious to
> > know if there is any of you with the same hardware who found a
> > workaround and would share it.
> 
> I do have a macbookpro and for me it was exactly the issue. e.g. The
> graphics adapter by default is not bus master.
> 
Sorry, I do not understand what #'is not bus master'# means

> > 
> > Looking at the various logs I can't find any clue. 
> > Reading  I also
> > consulted the suggested link 'http://askubuntu.com/a/613573/134848' but
> > as its author worked on a MacBook7,1 whereas mine is macmini4,1 I'm
 ~~
> > afraid either I misunderstood his advices concerning the 'setpci -s' (my
> > values where different) or that patch is not applicable to macmini.
> > 
> 
> 
> what is the result of 
> nvidia-detect
> ?
I can't tell because in the meantime I went back to the 'nouveau'
driver: up to now it hasn't yet frozen (cross fingers ;-)
However, Xorg.?.log showed the module was detected and loaded and this
fact is asserted by the following lines extracted from a 'journalctl -xb
--boot=-x' (where 'x' was the number of boot when nvidia was installed)

> 
> what does 
> sudo lshw -businfo -class bridge -class display
> report back?
> 

Ok, this is the result with nouveau installed, but it was identical when
nvidia was installed (of course, being hw stuff)

Bus info  Device  Class  Description

pci@:00:00.0  bridge MCP89 HOST Bridge
pci@:00:03.0  bridge MCP89 LPC Bridge
pci@:00:0e.0  bridge NVIDIA Corporation
pci@:01:00.0  bridge XIO2213A/B/XIO2221 PCI Express to 
PCI Bridge [Cheetah Express]
pci@:00:15.0  bridge NVIDIA Corporation
pci@:00:16.0  bridge NVIDIA Corporation
pci@:00:17.0<->   bridge MCP89 PCI Express Bridge
pci@:05:00.0<->   displayMCP89 [GeForce 320M]

> and the file you created
> /etc/grub.d/01_enable_vga.conf

> can you pleas list the contents of this file?

Here it is (I saved it before removng nvidia):

# 01_enable_vga.conf
# per modificare i valori del bridge e del display della nvidia
# cfr. proprietary-nvidia-drivers-on-mac-with-efi (ask ubuntu)
  setpci -s "00:17.0"  3e.b=8
  setpci -s "05:00.0"  04.b=7

>
Note that when I checked, as suggested by the ubuntu link author, with
  $ sudo setpci -s "00:17.0" 3e.b
08
  $ sudo setpci -s "05:00.0" 04.b
07
I obtained the correct values.

Some further considrations:
While I'm writing this answer I'm recalling to my mind that I was acting
in complete ignorance of what all that mean, particularly the 3e.b=8
etc. In the meantime I have desperately searched the web for an
explanation of hex structure, which I found a few hours ago. And the
only doubt, now, is that I might have first given that setpci
instruction with the same values of ubuntu post (05.b=7) and after
getting the black screen having corrected to 04.b=7.
I am still not sure which value would have been correct and if you know
please tell me just for the case I might go back again to nvidia in case
of new frequent frozes of nouveau.

Many thanks for your help. Regards, Ennio

-- 
[Perche' usare Win$ozz (dico io) se ..."anche uno sciocco sa farlo.   \\?//
 Fa' qualche cosa di cui non sei capace!"  (diceva Henry Miller) ](°|°)
[Why use Win$ozz (I say) if ... "even a fool can do that.  )=(
 Do something you aren't good at!" (as Henry Miller used to say) ]

Re: Server che non rilascia la ram

[Davide Prina]

On 11/03/2018 12:58, Michele Orsenigo wrote:

On domenica 11 marzo 2018 11:39:51 CET Davide Prina wrote:



Sul server hai installato anche gnome?


E' un server "puro": nessun ambiente grafico, non è installato neppure xorg


quindi magari anche il problema che ho segnalato io non è di gnome, ma 
di qualche altra componente, magari Linux stesso...


La cosa strana è che quasi nessuno ha notato qualcosa di simile... 
almeno io nelle ricerche che ho fatto non ho trovato segnalazioni.


Ciao
Davide

--
Dizionari: http://linguistico.sourceforge.net/wiki
I didn't use Microsoft machines when I was in my operational phase, 
because I couldn't trust them.
Not because I knew that there was a particular back door or anything 
like that, but because I couldn't be sure.

Edward Snowden

Re: [Debian-Stretch on macmini-4.1] Can't get X working with nvidia

[Henning Follmann]

On Sat, Mar 10, 2018 at 12:21:17AM +0100, Ennio-Sr wrote:
> Hi all!
> 
> since I installed new kernels (i.e. 4.9.0 and 4.14.0-0) to be
> half-protected against Meltdown/Spectre I was having freezing problems
> with X using 'nouveau' driver (never had any problem with older
> kernels). So i decided to try the original non-free nvidia driver as
> provided by the nvidia-legacy-340xx-driver and relted packages. Now,
> after struggling for a couple of days to search the web for a possible
> solution to the 'black screen' coming out running startx, I'm curious to
> know if there is any of you with the same hardware who found a
> workaround and would share it.

I do have a macbookpro and for me it was exactly the issue. e.g. The
graphics adapter by default is not bus master.

> 
> Looking at the various logs I can't find any clue. 
> Reading  I also
> consulted the suggested link 'http://askubuntu.com/a/613573/134848' but
> as its author worked on a MacBook7,1 whereas mine is macmini4,1 I'm
> afraid either I misunderstood his advices concerning the 'setpci -s' (my
> values where different) or that patch is not applicable to macmini.
> 


what is the result of 
nvidia-detect
?


what does 
sudo lshw -businfo -class bridge -class display
report back?

and the file you created
/etc/grub.d/01_enable_vga.conf

can you pleas list the contents of this file?




-- 
Henning Follmann   | hfollm...@itcfollmann.com

Re: Server che non rilascia la ram

[Michele Orsenigo]

On domenica 11 marzo 2018 22:33:38 CET Felipe Salvador wrote:
> > > Io riporterei il valore di swappiness a quello di default (60 se non
> > > ricordo male).
> > > 
> > > Leggi qui https://www.linuxatemyram.com/
> > 
> > Ok, ci ripenserò sulla swap: avevo seguito un consiglio legato a mariadb.
> > In ogni caso il problema esisteva anche prima di portare la swappiness a 0
> 
> Se vuoi approfondire oggi ho trovato questo articolo, molto
> interessante. È in inglese, il contenuto vale bene lo sforzo
> per tradurla.
> 
> https://chrisdown.name/2018/01/02/in-defence-of-swap.html
> 

Grazie me lo leggo.
E oggi mi studio anche il tuning di httpd

Buona giornata

-- 
Michele Orsenigo
deb...@orsenigo.it

Re: Server che non rilascia la ram

[Michele Orsenigo]

On domenica 11 marzo 2018 13:41:46 CET Alessandro Pellizzari wrote:
> On 11/03/18 12:10, Michele Orsenigo wrote:
> > I servizi che rallentano, fino a bloccarsi sono fondamentalmente mariadb e
> > dovecot.
> 
> Magari l'hai già fatto, ma prova a controllare la configurazione delle
> cache di MariaDB. La config suggerita da loro (20% della RAM con MyISAM
> e 70% con InnoDB) presuppone 4GB di RAM, ma consigliano di abbassarla su
> sistemi con meno RAM.
> 
> Nel tuo caso fai girare anche altra roba sullo stesso computer, quindi
> mi terrei ancora più basso, altrimenti rischi di "rubare" cache ad altri
> servizi, rendendoli molto lenti.
> 
> Vedi qui: https://mariadb.com/kb/en/library/mariadb-memory-allocation/
> 

Ho modificato la configurazione di mariadb seguendo l'indicazione che mi hai 
dato.
Vediamo che succede :)


-- 
Michele Orsenigo
deb...@orsenigo.it