IRC channel logs

2021-12-29.log

back to list of logs

<damo22>thanks for the patch biblio, i applied it to my tree but i didnt have your name and email so i just applied it as my user
<damo22>however, there is still work to be done for the PCI interrupt mapping
<biblio>damo22: sure.
<biblio>damo22: yes could you point some hits for interrupt mapping. I am still reading docs for HURD development.
<damo22>its in acpi_init.c, that link i gave you last time to a particular FIXME
<damo22>its actually not hurd related, its more in ACPI
<biblio>damo22: yes, I read some docs about ACPI.
<biblio>damo22: You mean locking part for acpica-nothread branch.
<damo22>no not the locking. I mean acpi_get_irq_...
<damo22>we need to fix that function so it works in all cases
<damo22>we can probably live with acpi being single threaded for now
<damo22>but we need it to return the correct irqs
<biblio>damo22: I can check fix the locking issue for acpi.
<damo22>sure but acpi is useless until it can deliver the irqs
<biblio>damo22: one thing i did not get. "TODO: Install interrupt handler for ACPI".
<damo22>im not sure about that either yet
<biblio>damo22: ok now I will try to make acpi_get_irq_number() work for all cases.
<biblio>damo22: it worked for my simple case. which cases are missing or I should focus on.
<damo22>see the FIXMEs
<biblio>damo22: ok for interrupt i can check linux part and get some idea for interrupt implementation for ACPI
<biblio>damo22: ok now I understand.
<biblio>damo22: gtg. i will update after making progress. thanks
<deavmi>What debian gnu/hurd image should I use
<deavmi>need one that works
<deavmi>I get driver issues in the installer, driver issues pertaining to removable media
<deavmi>I used an image suggested here a while back, some nigthyl, but I don't know where to find it
<deavmi>((
<Pellescours>deavmi: cf topic => disk image: https://tinyurl.com/yc5u7n9y
<deavmi>oh soryr
<deavmi>the whole topic didnd't display
<deavmi>thanks for that
<luckyluke>Hi! should the current 64-bit gnumach work with a simple user-space executable using only mach_print()?
<youpi>In my last tests the system calls were working, yes
<youpi>(with 32bit userland, that is)
<luckyluke>ok. is there already some work on adapting boothdr.S to 64 bit, for example on qemu? I was playing with it, but it seems it's not the only place that needs to be adapted.
<youpi>I don't remember somebody working on that
<luckyluke>for example the multiboot part, but that doesn't seem too difficult. I was also wondering about the initial page table map, why is VM_MIN_KERNEL_ADDRESS set to 0x40000000 and not to something higher, like 0xffffffff80000000 as suggested by mcmodel=kernel?
<Pellescours>luckyluke: I tried a bit but I don’t have really experience on that
<luckyluke>I was also looking at x15, which uses a separate boot section in the linker script
<luckyluke>but maybe creating a minimal page table map before jumping to C code would be enough
<youpi>luckyluke: I don't remember, I guess that was to make it simpler for now with 32/64 bit oddities
<youpi>luckyluke: is x86_64/boothdr.S not enough for multiboot yet?
<luckyluke>youpi: for multiboot, the issue is that in some places the struct definition is wrong when compiling for 64 bit
<youpi>patch welcome :)
<ArneBab>I now sent a patch with the checkperms translator patch and documentation to bug-hurd: https://mail.gnu.org/archive/html/bug-hurd/2021-12/msg00066.html