ChanServ changed the topic of #armlinux to: ARM kernel talk [Upstream kernel, find your vendor forums for questions about their kernels] | https://libera.irclog.whitequark.org/armlinux
heat_ has joined #armlinux
heat has quit [Read error: Connection reset by peer]
heat_ is now known as heat
agraf has quit [Ping timeout: 246 seconds]
agraf has joined #armlinux
heat has quit [Ping timeout: 252 seconds]
sakman_ is now known as sakman
haritzondo has joined #armlinux
haritz has quit [Read error: Connection reset by peer]
cbeznea_ has joined #armlinux
manchaw has joined #armlinux
monstr has joined #armlinux
monstr has quit [Ping timeout: 245 seconds]
houze has quit [Ping timeout: 246 seconds]
houze has joined #armlinux
jclsn has joined #armlinux
frieder has joined #armlinux
pivi has quit [Server closed connection]
pivi has joined #armlinux
monstr has joined #armlinux
mripard has joined #armlinux
rgallaispou has quit [Ping timeout: 258 seconds]
rgallaispou has joined #armlinux
ezulian has joined #armlinux
prabhakarlad has joined #armlinux
rgallaispou has left #armlinux [#armlinux]
smoothdev has quit [Quit: smoothdev]
smoothdev has joined #armlinux
headless has joined #armlinux
sszy has joined #armlinux
nsaenz has joined #armlinux
smoothdev has quit [Quit: smoothdev]
smoothdev has joined #armlinux
smoothdev has quit [Client Quit]
smoothdev has joined #armlinux
ezulian has quit [Quit: ezulian]
ezulian has joined #armlinux
nsaenz has quit [Remote host closed the connection]
smoothdev has quit [Quit: smoothdev]
lag has quit [Ping timeout: 255 seconds]
HdkR has quit [Server closed connection]
HdkR has joined #armlinux
nsaenz has joined #armlinux
nsaenz has quit [Ping timeout: 252 seconds]
Amit_T has joined #armlinux
lag has joined #armlinux
nsaenz has joined #armlinux
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
nsaenz has quit [Remote host closed the connection]
headless has quit [Quit: Konversation terminated!]
<NishanthMenon>
Does anyone know the magic sauce for getting JTAG + software breakpoints to work with kernel? I mean, I did disable CONFIG_ARM_CORESIGHT_PMU_ARCH_SYSTEM_PMUCONFIG_CORESIGHT -> but ran headlong into this little problem: https://gist.github.com/nmenon/96a0122c82a3a11452ca152dba1e7786 - openocd s/w breakpoints dont work since the text areas are ro - h/w breakpoints work.. but are limited in number (asking for someone else)
<jn>
NishanthMenon: you can make the .text sections RW by disabling CONFIG_STRICT_KERNEL_RWX
<NishanthMenon>
aah thanks
<NishanthMenon>
trying..
nsaenz has joined #armlinux
benh_ has quit [Server closed connection]
benh has joined #armlinux
<robmur01>
NishanthMenon: IIRC the arguments for minimising hardware debug frustration are "rodata=off cpuidle.off=1"
<NishanthMenon>
Thanks robmur01 - that is much easier to try.. adding it to nokaslr
<NishanthMenon>
robmur01: worked like a charm. thank you!
<robmur01>
(disabling idle is because historically Arm Development Studio didn't really like CPUs suddenly disappearing - dunno if it's got any better since I last tried)
<NishanthMenon>
robmur01: ok the only thing irky is coresight driver taking over coresight regs and messing with openocd control :( -> will see if there is command line option for that sometime..
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
nsaenz has quit [Remote host closed the connection]
haritzondo is now known as haritz
haritz has quit [Changing host]
haritz has joined #armlinux
nsaenz has joined #armlinux
nsaenz has quit [Remote host closed the connection]
Amit_T has quit [Ping timeout: 258 seconds]
heat has joined #armlinux
smoothdev has joined #armlinux
smoothdev has quit [Client Quit]
Amit_T has joined #armlinux
nsaenz has joined #armlinux
nsaenz_ has joined #armlinux
nsaenz has quit [Read error: Connection reset by peer]
nsaenz_ has quit [Read error: Connection reset by peer]
ezulian has joined #armlinux
nsaenz has quit [Ping timeout: 255 seconds]
frieder has quit [Remote host closed the connection]
jwerner has quit [Server closed connection]
jwerner has joined #armlinux
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #armlinux
headless has joined #armlinux
alpernebbi has quit [Ping timeout: 255 seconds]
prabhakarlad has joined #armlinux
alpernebbi has joined #armlinux
prabhakarlad has quit [Quit: Client closed]
hanetzer has quit [Ping timeout: 252 seconds]
hanetzer has joined #armlinux
tlwoerner has quit [Server closed connection]
tlwoerner has joined #armlinux
mripard has quit [Quit: mripard]
<HdkR>
Is there a userspace accessible system register for enabling and disabling ASIMD execution? I have an edge case which would be really nice to have lazy FP state saving and restoring.
headless has quit [Ping timeout: 255 seconds]
hanetzer has quit [Ping timeout: 248 seconds]
<Bitweasil>
Don't you get that by default on the Linux kernel these days?
headless has joined #armlinux
<Bitweasil>
I thought it left that stuff disabled on context switch until you fault, at which point it swaps state around.
<Bitweasil>
Though the microarchitectural leakages from it may have been responsible for some changes there, I've not kept up.
<j`ey>
aarch64 linux doesn't do lazy siwtching for ASIMD
<maz>
j`ey: not for userspace. but we do it for guests (and even that isn't that lazy).
Amit_T has quit [Remote host closed the connection]
prabhakarlad has joined #armlinux
cbeznea_ has joined #armlinux
<HdkR>
maz: Dang, So I can't do it in userspace at all. Good to know.
<HdkR>
I think I would have needed a CPACR_EL0 or something?
<j`ey>
yeah that controls the trapping
<HdkR>
Guess I'll just need to think of a smarter way to not need to loadstore all this state :D
headless has quit [Quit: Konversation terminated!]
<maz>
HdkR: the problem with having a CPACR_EL0 is that you need to route the exception to... EL0. and we don't have userspace exceptions. the best you could have is some ugly API with the kernel to disable FP for you and give you a signal (+ re-enabling FP) on FP execution.
<HdkR>
Which would end up being worse than just saving and restoring my state once I need to touch the kernel :)
<maz>
exactly.
heat has quit [Remote host closed the connection]
heat has joined #armlinux
<broonie>
HdkR: OOI what’s the use case?
<HdkR>
broonie: I need to save and restore 30 FPRs when jumping out of some JIT code and coming back. Which is something like 11 instructions on either side of the call and 480 bytes of memory access. Trimming that down to a theoretical MRS/MSR plus masking would be a big performance win
<HdkR>
Since the call out of code is unlikely to touch FPRs anyway