Re: Fwd: Comments about SMP

2019-04-15 Thread Almudena Garcia
Hi other time: A week ago, I got to implement the cpu enabling during gnumach boot. But the cpus aren't available to the system yet, I have to configure their Local APIC, and add them to the scheduler. But I don't know how can I add the cpus to the scheduler. I've seen some functions, as *sched_i

Fwd: Comments about SMP

2019-04-02 Thread Almudena Garcia
I forgot to resend all -- Forwarded message - De: Almudena Garcia Date: mar., 2 abr. 2019 a las 14:31 Subject: Re: Comments about SMP To: Hi all: I've got to implements the cpu enumeration reading ACPI tables, and to boot the system with *mach_ncpus=2* and a only cpu. &

Re: Comments about SMP

2019-03-10 Thread Almudena Garcia
>> It appears mach_ncpus is hardcoded at compile time, maybe you just need > to detect this value via ACPI (?) Yes, this is the idea. I set mach_ncpus to maximum value to avoid broken the cpu structure arrays. But It's not the only that I need to enable cpus. I also need to find their APIC s

Re: Comments about SMP

2019-03-10 Thread Damien Zammit
Hi Almudena, > El dom., 10 mar. 2019 a las 18:35, Samuel Thibault > (mailto:samuel.thiba...@gnu.org>>) escribió: > Adam Van Ymeren, le dim. 10 mars 2019 13:08:23 -0400, a ecrit: > > I don't think that's necessary.  The process doesn't have to > initiated from gnumach.  Hurd could have

Re: Comments about SMP

2019-03-10 Thread Almudena Garcia
Ok. I'll take notes about this idea. Currently, I have an implementation to find lapic and ioapic from ACPI tables, from the kernel. In the test model runs correctly, now I'm porting this to gnumach. This is a preliminary implementation: https://github.com/AlmuHS/GNUMach_SMP/blob/wip/kern/acpi_rs

Re: Comments about SMP

2019-03-10 Thread Samuel Thibault
Adam Van Ymeren, le dim. 10 mars 2019 13:08:23 -0400, a ecrit: > On March 10, 2019 11:58:14 AM EDT, Almudena Garcia > wrote: > >> My point is that it doesn't have to be during boot, it could be after > >> userland started a least a bit. > >> > > > >A simple question: Is It possible to run a routi

Re: Comments about SMP

2019-03-10 Thread Adam Van Ymeren
On March 10, 2019 11:58:14 AM EDT, Almudena Garcia wrote: >> >> My point is that it doesn't have to be during boot, it could be after >> userland started a least a bit. >> > >A simple question: Is It possible to run a routine in userland from >gnumach? >Or It's necessary to run a external user

Re: Comments about SMP

2019-03-10 Thread Almudena Garcia
> > My point is that it doesn't have to be during boot, it could be after > userland started a least a bit. > A simple question: Is It possible to run a routine in userland from gnumach? Or It's necessary to run a external userland application, as a Hurd server. I don't know if could be possible

Re: Comments about SMP

2019-03-10 Thread Samuel Thibault
Almudena Garcia, le dim. 10 mars 2019 16:04:29 +0100, a ecrit: > I don't sure about you says.  Are you suggesting to boot the other cores from > a > Hurd server? I mean to possibly provide from userland to the kernel the information for booting the other cores. > My idea is about to implement SM

Re: Comments about SMP

2019-03-10 Thread Almudena Garcia
I don't sure about you says. Are you suggesting to boot the other cores from a Hurd server? My idea is about to implement SMP support from gnumach, searching and starting the cpus during the boot. I can find the other cpus (the local apic and ioapic) from ACPI tables, but after this I need to sen

Re: Comments about SMP

2019-03-10 Thread Samuel Thibault
Almudena Garcia, le lun. 04 févr. 2019 12:08:28 +, a ecrit: > Nowadays, nobody uses MP Table, which is deprecated since 1997. Instead, > modern OS use ACPI tables. Thinking about it: parsing ACPI tables could be done in userland. Userland could then tell the kernel how to boot other cores. W

Re: Comments about SMP

2019-02-04 Thread Almudena Garcia
Nowadays, nobody uses MP Table, which is deprecated since 1997. Instead, modern OS use ACPI tables. The OSKIT SMP functions feels good: may I could use this as base. El Lunes 4 de febrero de 2019, Damien Zammit escribió: > Hi, > > On 04/02/19 06:27, Almudena Garcia wrote: > > Possibly. I

Re: Comments about SMP

2019-02-04 Thread Damien Zammit
Hi, On 04/02/19 06:27, Almudena Garcia wrote: > Possibly. I don't know if it's still relevant with nowadays' hardware, > though: is APIC still at that address? See [1] for coreboot's definition of the x86 multiprocessing table MP table structures. See [2] for coreboot's implementation of

Re: Comments about SMP

2019-02-03 Thread Almudena Garcia
> > though: is APIC still at that address? > I don't know, but we can search the current address reading from ACPI tables, so It's not problem. Another question: I found a project called OSKit, what feels to be Mach 2.x branch. What is the difference about gnumach and oskit? In their repository

Re: Comments about SMP

2019-02-03 Thread Samuel Thibault
Hello, Almudena Garcia, le dim. 03 févr. 2019 19:45:21 +0100, a ecrit: > At first time, I found that Mach 4 also had a implementation of cpu_number() > (in kernel/imps/cpu_number.h), with this: > > static inline int > cpu_number() > { >     return apic_local_unit.unit_id.r >> 24; > } > > Later,

Comments about SMP

2019-02-03 Thread Almudena Garcia
Hi all: I'm researching about SMP support in Mach microkernel, reading code and reviewing old messages in maillist. Mach 4 code can be downloaded from here: ftp://ftp.lip6.fr/pub/mach/mach4/mach/ And, exploring old code, I'm find some interesting things: At first time, I found that Mach 4 also