Re: Kernel-Crash when working with ubt0

From: <mms.vanbreukelingen_at_gmail.com>
Date: Tue, 27 Aug 2019 08:02:57 +0000 (UTC)
 
  On Tue, 27 Aug 2019 at 9:11, Konstantin Belousov<kostikbel_at_gmail.com> wrote:   On Mon, Aug 26, 2019 at 02:35:25PM -0700, maksim yevmenkin wrote:
> 
> 
> > On Aug 26, 2019, at 9:14 AM, Warner Losh <imp_at_bsdimp.com> wrote:
> > 
> > Is it from read_connection_list? If so I have a 'patch' that I'm using but
> > haven't committed because it's just too gross: drop the lock before the
> > copyout and pick it up again after...
> > 
> 
> Hmm... interesting.... 
> 
> I only took a brief look at it. I suppose I can ensure user space address is wired and then copyout() can be called with mutex held

>No, you cannot do this, at least without making the kernel to panic.
User might unmap the wired mapping at any time still.
Is it a userspace specific issue,  I don't think so?  It's a page fault because of aquired lock. 
Well,  I try two things this evening: 
1. I do the patch manually cut'n paste,  as git --diff is a weired thing,  especially when you come from Arch - but ok,  I give this a try. 
2. Recompile with option MTX_SPIN=y at the config at /usr/src/head/sys/x86_64/conf/MIRANDA, 
make with INSTKERNNAME=XYZ compiles ok,  but you just get an M at the reversion#, just the CURRENT appears with "uname -a". Would be a bit arrogant,  insulant and egoistic calling our Kernel by your first name and Helling everyone you've coded it,  IMO. 

Btw,  you can't just mix up LINT and "stock" together,  since I did,  I know what the long run of -1😭 means. Not a cool thing,  when compiling 8 hours at a (3rd,  to be honest)  world and then the kernel built just can't bootstrap.  
I also built two worlds 28 simultaniously from different sources and compared installation. mergemaster's -Ui (or -p) always's to be done at Single Mode with remounting root rw, system is very vulnerable at the moment masterpwd is constructed and groups are merged (this not in Single Mode as it needs the ELF64 for merging correctly (right?) and dbus missing some users afterwards normally. I had to recreate user not just once and to chown 775 it. 
It has something from an abortion when turning on after install Kernel and screen remains black. This is what all of US fear.  zcat under Linux can be a black cat...




> 
> It’s odd that I’ve never seen this before. I’m traveling at the moment will take a closer look this later this week
> 
> Thanks
> Max
> _______________________________________________
> freebsd-current_at_freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"
  

Received on Tue Aug 27 2019 - 06:03:02 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:21 UTC