Jarkko Sakkinen
Posts
4547Following
317Followers
476OpenPGP: 3AB05486C7752FE1
Jarkko Sakkinen
jarkko
I ended up with:
ftrace {
event.kprobes {
tpm_transmit {
filters = "common_pid < 2"
probes = "tpm_transmit"
hist {
keys = stacktrace
}
}
}
}
kernel {
trace_buf_size = 1M
trace_options = sym-addr
}
It gives me counts of stack exercised during the boot. I’d like to get latency calculation but I’m not really good at this so this can do for the time being.
The is held in /sys/kernel/debug/tracing/events/kprobes/tpm_end/hist.
0
0
0
Jarkko Sakkinen
jarkko
Edited 10 months ago
I voted in the Linux Foundation TAB elections.
Since it is politics I can tell that my top three was Dan Williams, Dave Hansen and Kees Cook and for the rest I gave equal vote.
Since it is politics I can tell that my top three was Dan Williams, Dave Hansen and Kees Cook and for the rest I gave equal vote.
0
0
0
Post-poning hwrng registeration will not help and was a false conclusion .
I attached kprobes using boot-time config and these appear when init process is already running:
start_event: (tpm_transmit+0x4/0x4b0)
end_event: (tpm_transmit_cmd+0x33/0xc0 <- tpm_transmit)
start_event: (tpm_transmit+0x4/0x4b0)
end_event: (tpm_transmit_cmd+0x33/0xc0 <- tpm_transmit)
start_event: (tpm_transmit+0x4/0x4b0)
end_event: (tpm_transmit_cmd+0x33/0xc0 <- tpm_transmit)
These emit from tpm2_get_random() invoked by hwrng.
I guess 8.9 seconds is what we satisfy for the moment :-) I put out one more version where extras have been cut and performance fixes also have fixes tags as we most likely want these also to 6.10/6.11.
I attached kprobes using boot-time config and these appear when init process is already running:
start_event: (tpm_transmit+0x4/0x4b0)
end_event: (tpm_transmit_cmd+0x33/0xc0 <- tpm_transmit)
start_event: (tpm_transmit+0x4/0x4b0)
end_event: (tpm_transmit_cmd+0x33/0xc0 <- tpm_transmit)
start_event: (tpm_transmit+0x4/0x4b0)
end_event: (tpm_transmit_cmd+0x33/0xc0 <- tpm_transmit)
These emit from tpm2_get_random() invoked by hwrng.
I guess 8.9 seconds is what we satisfy for the moment :-) I put out one more version where extras have been cut and performance fixes also have fixes tags as we most likely want these also to 6.10/6.11.
0
0
0
@popey @timojyrinki @ubuntuasahi I'm waiting the ARM PC market further mature but over time ARM64 laptop with UEFI/ACPI and TPM will be quite nice option because then you can encrypt the hard drive with TPM. UEFI and ACPI are fat and ugly but also super robust. I'd say Macbook Pro is right now quite feasible laptop if you don't do kernel development but for a kernel developer it is limited as hell.
0
0
1
@popey @timojyrinki @ubuntuasahi Still after using Macbook Pro for 1.5 years I actually prefer stock Thinkpads over it for the sake of robustness. I had to pick yesterday for my new job between Mac and PC laptop, and went with latter.
1
0
0
Jarkko Sakkinen
jarkko
@mattdm Im a recent Fedora user. I used 2004-2021 Ubuntu until it became trainwreck. After trying bunch of options Fedora delivers me pretty much I want: no bullshit desktop with desktop and server images made friendly for engineering other stuff expect the distribution itself 🙂 Congrats!
0
0
1
Jarkko Sakkinen
repeated
repeated
Matthew Miller
mattdm@hachyderm.ioWow! Fedora Linux 41 beta today, _and_ twelve years at Red Hat for me!
2
3
1
Jarkko Sakkinen
jarkko
Edited 11 months ago
After having MacBook Pro for 1.5 years from Tampere University picking the dev laptop for the new job was a no brainer options to pick being Macbook Pro or x86 Linux laptop.
MacOS is pure garbage a development environment. If you get one there is no other ways to make usable except installing Asahi Linux. Working with toolchains sucks, virtualization sucks and conceptually containers do not in reality exist.
I'm not even joking when I say that if i had picked Macbook Pro as my dev laptop for the new job it would statistically show in the performance reports over period of time as a negative impact. It just is super bad operating system for writing code.
I still love my personal Mac mini which is over-expensive audio plugin host :-) It does deliver in that task tho and works well also for watching Netflix.
Phew, it was nice to let all the hate for Macbook Pro out ;-) Piece of shit laptop tbh.
MacOS is pure garbage a development environment. If you get one there is no other ways to make usable except installing Asahi Linux. Working with toolchains sucks, virtualization sucks and conceptually containers do not in reality exist.
I'm not even joking when I say that if i had picked Macbook Pro as my dev laptop for the new job it would statistically show in the performance reports over period of time as a negative impact. It just is super bad operating system for writing code.
I still love my personal Mac mini which is over-expensive audio plugin host :-) It does deliver in that task tho and works well also for watching Netflix.
Phew, it was nice to let all the hate for Macbook Pro out ;-) Piece of shit laptop tbh.
1
0
2
Jarkko Sakkinen
jarkko
I was looking for something that would be a bit like Maple and Matlab for working as calculator, plotting and stuff like that but not as involved as those tools. After trying bunch of options I realized https://julialang.org/ is pretty much size-fit of this niche.
0
0
1
1
0
0
Jarkko Sakkinen
jarkko
I like eBPF because it allows me to set conditions and measurement points so freely. A performance issue that took me some years ago two days to uncover takes now few hours. Improving tools brings so much more productivity than ChatGPT ever could.
0
0
2