alperak has quit [Quit: Connection closed for inactivity]
vagrantc has quit [Ping timeout: 265 seconds]
prabhakalad has quit [Ping timeout: 252 seconds]
<smaeul_>
bjdooks: what else would you do? there's no instruction that can write satp and jump to a new PC at the same time
smaeul_ is now known as smaeul
prabhakalad has joined #riscv
<jrtc27>
create a transitional page table that maps both the current (physical or identity-mapped virtual) PC and the desired virtual PC to the current page, switch to that, invalidate mappings, jump to the virtual PC and then continue setting up the rest of the desired mappings
<jrtc27>
but that's more code, and it's not a fast path
mtoy has quit [Ping timeout: 240 seconds]
mtoy has joined #riscv
dramforever[m]1 has joined #riscv
<dramforever[m]1>
what's wrong with stvec anyway
<dramforever[m]1>
xip?
bigbl4ckw0lf has quit [Quit: Ping timeout (120 seconds)]
bigbl4ckw0lf has joined #riscv
BootLayer has joined #riscv
balrog has quit [Ping timeout: 252 seconds]
balrog has joined #riscv
dobson has quit [Ping timeout: 276 seconds]
ruidx has quit [Quit: WeeChat 3.8]
dobson has joined #riscv
vagrantc has joined #riscv
vagrantc has quit [Ping timeout: 276 seconds]
BootLayer has quit [Quit: Leaving]
alperak has joined #riscv
naoki has joined #riscv
agentcasey_ has quit [Quit: ZNC 1.10.x-git-82-9ab81d1d - https://znc.in]
agentcasey has joined #riscv
theChn has joined #riscv
ruidx has joined #riscv
cleger has joined #riscv
matrixbrain has joined #riscv
jjido has joined #riscv
matrixbrain has quit [Quit: Leaving]
Guest3594 is now known as chripo
IndyJoenz has quit [Ping timeout: 245 seconds]
IndyJoenz has joined #riscv
alexghiti has joined #riscv
ldevulder has joined #riscv
jjido has quit [Quit: My laptop has gone to sleep. ZZZzzz…]
<bjdooks>
It does seem to work but very confusing when it soon after switches to a new table which turns out has a bad kernel mapping
<Esmil>
with that and Geert / Marc Zyngiers recent fix PCIe actually seems to work on my IcicleKit with 6.16-rc5
<conchuod>
Applied, with a squash I think.
<conchuod>
I think that whether or works or not depends on the device you have and where Linux allocates your buffers
<Esmil>
conchuod: applied for 6.17?
<conchuod>
I thought it was applied prior to 6.16, but maybe not then. Bjorn certainly said he has applied it and I thought he had done so and squashed it together
<Esmil>
ah, that's probably the problem
<conchuod>
I think though there's a bug in it that I asked Daire to send a patch for.
<conchuod>
I'll mark both down on my todo list
<Esmil>
thanks!
<bjdooks>
anyone tried building a riscv kernel with gcc15 ?
<Esmil>
archlinux' cross-compiler says riscv64-linux-gnu-gcc (GCC) 15.1.0 and that seems to work fine
naoki has quit [Quit: naoki]
<bjdooks>
ok, i think uboot is at fault, but not sure why
<bjdooks>
i can boot qemu,virt with this kernel but not cva6's uboot
<bjdooks>
now it just hangs in udelay()
<geertu>
bjdooks: cycle counter not running?
<bjdooks>
geertu: will check later
hmw has quit [Quit: Bye.]
hmw has joined #riscv
prabhakalad has quit [Read error: Connection reset by peer]
<shadows>
la_mettrie: Yes or they should, same specification StarFive JH7110 CPU. There was a change to adopt upstream Linux devicetree in U-Boot, and then corresponding change in Linux to adopt the missing clock stuff from U-Boot to make that happen
<shadows>
la_mettrie: I don't understand the technical details so if you are more familiar with clocks and what that all should be please let us have this discussion to improve the situation :) I wish that there were better divisions of CPU clock rates but the people for StarFive vendor did not seem to understand what I am asking
<shadows>
there is currently a problem where the CPU runs at increased voltage and clock rate which is too much heat