Re: [PATCH for 6.2 38/49] bsd-user: Update mapping to handle reserved and starting conditions

[Warner Losh <imp@bsdimp.com>]

[-- Attachment #1: Type: text/plain, Size: 2721 bytes --]

On Tue, Aug 10, 2021 at 10:27 AM Richard Henderson <
richard.henderson@linaro.org> wrote:

> On 8/7/21 11:42 AM, Warner Losh wrote:
> > From: Warner Losh <imp@FreeBSD.org>
> >
> > Update the reserved base based on what platform we're on, as well as the
> > start of the mmap range. Update routines that find va ranges to interact
> > with the reserved ranges as well as properly align the mapping (this is
> > especially important for targets whose page size does not match the
> > host's). Loop where appropriate when the initial address space offered
> > by mmap does not meet the contraints.
> >
> > Signed-off-by: Mikaël Urankar <mikael.urankar@gmail.com>
> > Signed-off-by: Stacey Son <sson@FreeBSD.org>
> > Signed-off-by: Warner Losh <imp@bsdimp.com>
> > ---
> >   bsd-user/main.c |  23 ++-
> >   bsd-user/mmap.c | 372 ++++++++++++++++++++++++++++++++++++++++--------
> >   bsd-user/qemu.h |   5 +-
> >   3 files changed, 335 insertions(+), 65 deletions(-)
> >
> > diff --git a/bsd-user/main.c b/bsd-user/main.c
> > index 93ef9298b8..36852604f8 100644
> > --- a/bsd-user/main.c
> > +++ b/bsd-user/main.c
> > @@ -49,12 +49,29 @@
> >   #include "target_arch_cpu.h"
> >
> >   int singlestep;
> > -unsigned long mmap_min_addr;
> >   uintptr_t guest_base;
> >   static const char *cpu_model;
> >   static const char *cpu_type;
> >   bool have_guest_base;
> > +#if (TARGET_LONG_BITS == 32) && (HOST_LONG_BITS == 64)
> > +/*
> > + * When running 32-on-64 we should make sure we can fit all of the
> possible
> > + * guest address space into a contiguous chunk of virtual host memory.
> > + *
> > + * This way we will never overlap with our own libraries or binaries or
> stack
> > + * or anything else that QEMU maps.
> > + */
> > +# ifdef TARGET_MIPS
> > +/* MIPS only supports 31 bits of virtual address space for user space */
> > +unsigned long reserved_va = 0x77000000;
> > +# elif defined(TARGET_PPC64)
> > +unsigned long reserved_va = 0xfffff000;
> > +# else
> > +unsigned long reserved_va = 0xf7000000;
> > +# endif
> > +#else
> >   unsigned long reserved_va;
> > +#endif
>
> All of these 7's look to be copying an old linux-user bug.
> I cleaned this up in 18e80c55bb6.
>

Oh nice. I'd missed that one... a bit before I started hacking on bsd-user,
but quite nice indeed.


> Otherwise,
> Acked-by: Richard Henderson <richard.henderson@linaro.org>
>
> I would hope one day this memory management could be shared between the
> user-only
> implementations.  It's complicated enough...
>

Honestly, one reason I want to get "caught up" with upstream is for such a
refactor. There's
so much duplication that it's hard to keep up.

Warner

[-- Attachment #2: Type: text/html, Size: 3810 bytes --]