Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 26 Oct 2006 11:38:24 -0400
From:      John Baldwin <jhb@freebsd.org>
To:        Ruslan Ermilov <ru@freebsd.org>
Cc:        freebsd-stable@freebsd.org, Stefan Bethke <stb@lassitu.de>, Bruce Evans <bde@zeta.org.au>, Bruce Evans <bde@freebsd.org>
Subject:   Re: Still possible to directly boot without loader?
Message-ID:  <200610261138.24939.jhb@freebsd.org>
In-Reply-To: <20061026144240.GB1219@rambler-co.ru>
References:  <3A8131D4-881E-4873-A682-543A1A88C063@lassitu.de> <200610261028.10680.jhb@freebsd.org> <20061026144240.GB1219@rambler-co.ru>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday 26 October 2006 10:42, Ruslan Ermilov wrote:
> On Thu, Oct 26, 2006 at 10:28:09AM -0400, John Baldwin wrote:
> > On Thursday 26 October 2006 10:18, Ruslan Ermilov wrote:
> > > On Thu, Oct 26, 2006 at 10:52:30PM +1000, Bruce Evans wrote:
> > > > On Thu, 26 Oct 2006, Ruslan Ermilov wrote:
> > > > >3)  It's currently broken even on i386; backing out
> > > > >   rev. 1.71 of boot2.c by jhb@ fixes this for me.
> > > > >
> > > > >: revision 1.71
> > > > >: date: 2004/09/18 02:07:00;  author: jhb;  state: Exp;  lines: +3 -3
> > > > >: A long, long time ago in a CVS branch far away (specifically, HEAD prior
> > > > >: to 4.0 and RELENG_3), the BTX mini-kernel used paging rather than flat
> > > > >: mode and clients were limited to a virtual address space of 16 megabytes.
> > > > >: Because of this limitation, boot2 silently masked all physical addresses
> > > > >: in any binaries it loaded so that they were always loaded into the first
> > > > >: 16 Meg.  Since BTX no longer has this limitation (and hasn't for a long
> > > > >: time), remove the masking from boot2.  This allows boot2 to load kernels
> > > > >: larger than about 12 to 14 meg (12 for non-PAE, 14 for PAE).
> > > > >:
> > > > >: Submitted by:   Sergey Lyubka devnull at uptsoft dot com
> > > > >: MFC after:      1 month
> > > > 
> > > > The kernel is linked at 0xc0000000 but loade din low memory, so the high
> > > > bits must be masked off like they used to be for the kernel to boot at all.
> > > > This has nothing to do with paging AFAIK.  Rev.1.71 makes no sense, since
> > > > BTX isn't large, and large kernels are more unbootable than before with
> > > > 1.71.
> > > > 
> > > The real purpose of this commit was to allow to directly "load kernels
> > > larger than about 12 to 14 meg (12 for non-PAE, 14 for PAE)".  (Old
> > > version masked high 8 bits, leaving only 2^24=16MB for the kernel.)
> > > 
> > > I have compiled GENERIC and PAE kernels; objdump(1) reports that GENERIC
> > > kernel has virtual "start address 0xc0449cb0", and PAE has virtual "start
> > > address 0xc02458f0".
> > 
> > Yes, KERNLOAD for PAE is 2MB and for non-PAE is 4MB (to skip PSE page 0).
> > 
> > > What happens here is that BTX now uses flat memory model, and by not
> > > masking higher bits at all, BTX attempts to load kernels at above 3G,
> > > which silently fails, and then jumps to the entry point located in
> > > "no memory" unless the machine has enough memory.
> > > 
> > > If the machine has enough physical memory, e.g. 4G, then it works (I
> > > think that was the case on the machine John tested this change), but
> > > on my test machine I only have 3G of memory, so it fails.
> > 
> > Actually, it should never work, as the kernel assumes it is loaded at
> > KERNLOAD.
> > 
> > > My interim solution to the problem that would still allow booting
> > > larger than 16MB kernels is to mask some of the higher bits.
> > > Currently, I mask 28 bits that gives possible 256MB which is probably
> > > practical.
> > 
> > boot2 should do whatever loader does.
> > 
> But this would be a regression, since loader(8) does the following,
> in the ELF32 case:
> 
> : 0 edoofus:ttyp2:/sys/boot/i386/libi386 >grep -w entry elf32_freebsd.c
> :     vm_offset_t                 entry, bootinfop, modulep, kernend;
> :     entry = ehdr->e_entry & 0xffffff;
> :     printf("Start @ 0x%lx ...\n", entry);
> :     __exec((void *)entry, boothowto, bootdev, 0, 0, 0, bootinfop, modulep, kernend);

Ah, ok.  Make them both just mask the top 8 bits then. :)

-- 
John Baldwin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200610261138.24939.jhb>