Re: [RFC 2/3] lockdep: be nice about compiling from userspace

2012-10-25 Thread Sasha Levin
On 10/25/2012 01:06 PM, Ingo Molnar wrote: > * Sasha Levin wrote: >> Yes, we can wrap both static_obj() and debug_show_all_locks() >> with #ifndefs and let kvmtool provide it's own version of >> those two. > > Only static_obj() - I see no immediate reason why you shouldn't > be able to utilize

Re: [RFC 2/3] lockdep: be nice about compiling from userspace

2012-10-25 Thread Ingo Molnar
* Sasha Levin wrote: > On 10/25/2012 04:05 AM, Ingo Molnar wrote: > > > > * Sasha Levin wrote: > > > >> We can rather easily make lockdep work from userspace, although 3 issues > >> remain which I'm not sure about: > >> > >> - Kernel naming - we can just wrap init_utsname() to return kvmtool

Re: [RFC 2/3] lockdep: be nice about compiling from userspace

2012-10-25 Thread Sasha Levin
On 10/25/2012 04:05 AM, Ingo Molnar wrote: > > * Sasha Levin wrote: > >> We can rather easily make lockdep work from userspace, although 3 issues >> remain which I'm not sure about: >> >> - Kernel naming - we can just wrap init_utsname() to return kvmtool related >> utsname, is that what we wan

Re: [RFC 2/3] lockdep: be nice about compiling from userspace

2012-10-25 Thread Ingo Molnar
* Sasha Levin wrote: > We can rather easily make lockdep work from userspace, although 3 issues > remain which I'm not sure about: > > - Kernel naming - we can just wrap init_utsname() to return kvmtool related > utsname, is that what we want though? > > - static_obj() - I don't have a bette

[RFC 2/3] lockdep: be nice about compiling from userspace

2012-10-24 Thread Sasha Levin
We can rather easily make lockdep work from userspace, although 3 issues remain which I'm not sure about: - Kernel naming - we can just wrap init_utsname() to return kvmtool related utsname, is that what we want though? - static_obj() - I don't have a better idea than calling mprobe(), which so