Trifton has quit [Remote host closed the connection]
Trifton has joined #riscv
peepsalot has joined #riscv
BootLayer has joined #riscv
alperak has joined #riscv
mlw has joined #riscv
<xypron>
drewfustini: Concerning BPi-F3 on Ubuntu: The vendor kernel plus Ubuntu user land runs fine. Unfortunately the upstream kernel support for the board is not yet satisfying.
JRepin has quit [Quit: Konversation terminated!]
JRepin has joined #riscv
coldfeet has quit [Quit: Lost terminal]
JRepin has quit [Quit: Konversation terminated!]
JRepin has joined #riscv
hwspeedy has joined #riscv
Guest917 is now known as chripo
dramforever has joined #riscv
prabhakalad has quit [Ping timeout: 265 seconds]
prabhakalad has joined #riscv
ganboing has joined #riscv
<ganboing>
JohnHenry, you can use firmware events such as fw_sfence_vma_sent
<ganboing>
E.g., perf stat -a -A -e fw_sfence_vma_sent
<ganboing>
TLB shootdown is typically done by calling SBI firmware services (to do it more efficiently)
cleger has joined #riscv
ldevulder has joined #riscv
jobol has joined #riscv
<mps>
xypron: what version is vendor kernel
prabhakalad has quit [Quit: Konversation terminated!]
hwspeedy has quit [Read error: Connection reset by peer]
ruidx has joined #riscv
prabhakalad has joined #riscv
jacklsw has joined #riscv
jacklsw has quit [Quit: Back to the real world]
alexghiti has joined #riscv
Andre_Z has joined #riscv
Armand|X13s has joined #riscv
Armand|X13s has quit [Remote host closed the connection]
Armand|X13s has joined #riscv
Armand|X13s has quit [Quit: Leaving]
ldevulder has quit [Quit: Leaving]
dramforever has quit [Remote host closed the connection]
ruidx has quit [Ping timeout: 272 seconds]
naoki has joined #riscv
ldevulder has joined #riscv
JRepin has quit [Remote host closed the connection]
JRepin has joined #riscv
naoki has quit [Ping timeout: 260 seconds]
Andre_Z has quit [Quit: Leaving.]
ldevulder has quit [Quit: Leaving]
ldevulder has joined #riscv
jjido has joined #riscv
jjido has quit [Quit: My laptop has gone to sleep. ZZZzzz…]
mlw has quit [Ping timeout: 252 seconds]
mlw has joined #riscv
smaeul has joined #riscv
smaeul_ has quit [Ping timeout: 248 seconds]
ldevulder has quit [Ping timeout: 265 seconds]
<JohnHenry>
ganboing: Thank you, that's helpful
memset has quit [Remote host closed the connection]
memset has joined #riscv
unnick has quit [Ping timeout: 276 seconds]
jjido has joined #riscv
cleger has quit [Ping timeout: 265 seconds]
memset has quit [Remote host closed the connection]
memset has joined #riscv
<kepstin>
mps: the bianbu kernel for the k1/m1 chips is currently based on 6.6.63
jjido has quit [Quit: My laptop has gone to sleep. ZZZzzz…]
memset has quit [Remote host closed the connection]
<drewfustini>
xypron: thanks!
memset has joined #riscv
KombuchaKip has quit [Ping timeout: 272 seconds]
cp- has joined #riscv
jjido has joined #riscv
memset has quit [Remote host closed the connection]
<JohnHenry>
ganboing: my `perf list` doesn't have any SFENCE-type events I can see. I've got a Milk-V Pioneer running their old Fedora 38. Do you have any more suggestions?
<JohnHenry>
I realized `sudo perf list` was important and that has `tlb:tlb_flush`, which is a start, but I still don't see what might count shootdown interrupts specifically unless I can somehow filter the IPIs to catch those
<JohnHenry>
FWIW I'm 90% sure my system is using Linux IPIs and not the SBI to perform TLB shoodowns
jjido has quit [Quit: My laptop has gone to sleep. ZZZzzz…]
<bjdooks>
could you also kprobe on the relevant tlb flush function?
<JohnHenry>
that would be great if I knew how to kprobe. :)
<JohnHenry>
kprobes allow arbitrary monitoring in the kernel, right?
<bjdooks>
mostly
<bjdooks>
there's a few places you just can't
<JohnHenry>
any tutorial suggestions?
jobol has quit [Quit: Leaving]
<JohnHenry>
I guess using Linux IPIs is why I'm not getting the SBI events, but I guess RISC-V Linux doesn't have corresponding tracepoint events for its own IPI shootdowns?
<palmer>
JohnHenry looks like those are x86-only? Unless I'm missing something. Seems reasonable to add them, though...
<JohnHenry>
That's what I saw and thought, I'm glad you think the same! Maybe that would be a good Baby's First Patch for someone like me?
<palmer>
ya, go for it. I think it should just be a matter of decorating some TLB flushing stuff with trace_tlb_flush(). Might even be an asm-generic thing, depending on how it's all hooked together
alexghiti has quit [Quit: Connection closed for inactivity]