netbsd-tech-kern mailing list

Fromsort iconSubjectDate
David Laight
Re: semantics of TRYEMULROOT
It is a lot less of a hack than what went before. The only other option would be an actual (working) union mount, and that has much bigger problems - not to mention needing process specific mount tables! OTOH I do wish the 32bit/64bit compatibility had been done as part of the '64bit' port - so that the 64bit userspace doesn't have to rename anything from the 32bit version - giving you a userpace that can (at least in theory) boot either a 32bit or 64bit kernel. David -- David ...
Jan 4, 12:04 pm 2011
Joerg Sonnenberger
__ps_strings and compat32
Hi all, while investigating ways to access the auxillary vector, I found that we currently don't deal correctly with struct ps_strings in 32bit compat. This structure is placed by the kernel into the initial stack and referenced in userland by __ps_strings. It is used by setproctitle(3) and therefore e.g. by programs that want to include a status report in the ps(1) output. This naturally works only by chance, if at all in compat32. How do we want to deal with it? Variant 1: Leave it as ...
Jan 4, 5:09 am 2011
Matt Thomas
Re: CVS commit: src/sys/uvm
Still on my TODO list. I've been preparing to merge the mips64 branch the my mpc85xx code. One thing I've been thinking about is how to use an MMU's capability to map >1 page with a single TLB entry or PTE. If we are going with vm_physseg as a fundamental VM data structure, maybe we should think it about how to use it a little more. should it have a struct vm_page *(*ps_off2pg)(struct vm_physseg *, u_long frame_no); struct vm_page_md *(*ps_off2md)(struct vm_physseg *, u_long ...
Jan 3, 11:28 pm 2011
Masao Uebayashi
Re: physical address management, round 2
First step would be to start with populating the single, MI "phys_space_arena" and manage physical address "space". (This needs vmem(9) to be bootstrapped very early.) If this is done, and with my vm_physseg'ification, lots of MD code would be cleaned up.
Jan 4, 8:49 am 2011
Masao Uebayashi
Re: CVS commit: src/sys/uvm
Do you have any questions yet? -- Masao Uebayashi / Tombi Inc. / Tel: +81-90-9141-4635
Jan 3, 9:53 pm 2011
Masao Uebayashi
Re: CVS commit: src/sys/uvm
I take silence as "no objection". -- Masao Uebayashi / Tombi Inc. / Tel: +81-90-9141-4635
Jan 3, 10:01 pm 2011
Masao Uebayashi
Re: semantics of TRYEMULROOT
IMO emul lookup is completely a hack, and it should be redesigned using per-process namespace. That would answer many questions including RUMP namespace and/or devfs. -- Masao Uebayashi / Tombi Inc. / Tel: +81-90-9141-4635
Jan 4, 8:01 am 2011
Masao Uebayashi
Re: CVS commit: src/sys/uvm
This would be easy for cdevs if vm_physseg is allocated; cdev device pager returns the vm_physseg back to the fault handler. UVM can know if the faulting page is continugous physically or virtually Typical vnode lookup done in fault handling is: vaddr_t -> vm_map_entry -> uobj -> vm_page -> vm_physseg -> vm_page_md As I already explained many times, you need only 2 kind of lookups: vm_page -> vm_physseg + off vm_physseg + off -> vm_page_md (pvh)
Jan 4, 7:57 am 2011
previous daytodaynext day
January 3, 2011January 4, 2011None