New kernel options to consider

H_TeXMeX_H · 06-27-2010, 06:15 AM

So, I'm building the 2.6.34 kernel and I notice some interesting options, which you may want to consider including in the build, if you are building it:

Code:

  ┌───────────────────── Reroute for broken boot IRQs ──────────────────────┐
  │ CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS:                                │  
  │                                                                         │  
  │ This option enables a workaround that fixes a source of                 │  
  │ spurious interrupts. This is recommended when threaded                  │  
  │ interrupt handling is used on systems where the generation of           │  
  │ superfluous "boot interrupts" cannot be disabled.                       │  

  │ Symbol: X86_REROUTE_FOR_BROKEN_BOOT_IRQS [=y]                           │  
  │ Prompt: Reroute for broken boot IRQs                                    │  
  │   Defined at arch/x86/Kconfig:830                                       │  
  │   Depends on: X86_IO_APIC [=y]                                          │  
  │   Location:                                                             │  
  │     -> Processor type and features                                      │

Code:

  ┌─────────── Low address space to protect from user allocation ───────────┐
  │ CONFIG_DEFAULT_MMAP_MIN_ADDR:                                           │  
  │                                                                         │  
  │ This is the portion of low virtual memory which should be protected     │  
  │ from userspace allocation.  Keeping a user from writing to low pages    │  
  │ can help reduce the impact of kernel NULL pointer bugs.                 │  

  │ This value can be changed after boot using the                          │  
  │ /proc/sys/vm/mmap_min_addr tunable.                                     │  
  │                                                                         │  
  │ Symbol: DEFAULT_MMAP_MIN_ADDR [=4096]                                   │  
  │ Prompt: Low address space to protect from user allocation               │  
  │   Defined at mm/Kconfig:226                                             │  
  │   Depends on: MMU [=y]                                                  │  
  │   Location:                                                             │  
  │     -> Processor type and features                                      │

I know I've a seen thread about this one, kernel NULL pointer bugs.

H_TeXMeX_H · 06-27-2010, 08:40 AM

Oh, and I forgot to mention why I compiled a new kernel, it's because my USB flash drive. Whenever I try to copy large files to it, the transfer rate is very very slow, like under 1 MB/s. So, I read that a new kernel may fix it, and it does, I just tested it and the bug is fixed with the latest kernel, transfer rate is now over 8MB/s even with large files, dd reports 12.4 MB/s.

I think this is the option that fixes it:

Code:

  ┌─────── Improved Transaction Translator scheduling (EXPERIMENTAL) ───────┐
  │ CONFIG_USB_EHCI_TT_NEWSCHED:                                            │  
  │                                                                         │  
  │ This changes the periodic scheduling code to fill more of the low       │  
  │ and full speed bandwidth available from the Transaction Translator      │  
  │ (TT) in USB 2.0 hubs.  Without this, only one transfer will be          │  
  │ issued in each microframe, significantly reducing the number of         │  
  │ periodic low/fullspeed transfers possible.                              │  
  │                                                                         │  
  │ If you have multiple periodic low/fullspeed devices connected to a      │  
  │ highspeed USB hub which is connected to a highspeed USB Host            │  
  │ Controller, and some of those devices will not work correctly           │  
  │ (possibly due to "ENOSPC" or "-28" errors), say Y.                      │  
  │                                                                         │  
  │ If unsure, say N.                                                       │  
  │                                                                         │  
  │ Symbol: USB_EHCI_TT_NEWSCHED [=y]                                       │  
  │ Prompt: Improved Transaction Translator scheduling (EXPERIMENTAL)       │  
  │   Defined at drivers/usb/host/Kconfig:74                                │  
  │   Depends on: USB_SUPPORT [=y] && USB_EHCI_HCD [=m] && EXPERIMENTAL [=y │  
  │   Location:                                                             │  
  │     -> Device Drivers                                                   │  
  │       -> USB support (USB_SUPPORT [=y])                                 │  
  │         -> Support for Host-side USB (USB [=y])                         │  
  │           -> EHCI HCD (USB 2.0) support (USB_EHCI_HCD [=m])             │

linus72 · 06-27-2010, 08:55 AM

maybe not at all related H_TeXMeX_H
but, I compiled new zen kernel from packer

packer -S kernel26-zen

and now xconf; used in my arch/slack nfluxos builds wont work
xorg-server is 1.8.1

with Quax's 2.6.33.3 kernel it all works....

do you have any ideas?
sorry to hijack your thread, but my kernel under General says

[ ] Automatically SCHED_ISO policy for X

and I didn't "check" it[*]
could that be an issue?
what does that mean- "[ ] Automatically SCHED_ISO policy for X"

I noticed many new things with this kernel and dont know what they are?
This is also the original kernel I had the issue with x86_64 arch as we discussed in that thread...

H_TeXMeX_H · 06-27-2010, 09:43 AM

I'm pretty sure it is optional, so you don't need it.

You know, if you're compiling a kernel, you may want to look through this:
http://www.kroah.com/lkn/
see chapter 7 especially.

I usually start with the .config that comes with the distro, then remove and add modules as necessary. You kinda have to keep trying and thinking about it until you get it right. It takes a while.