* [Qemu-devel] target_phys_addr_t definition
@ 2009-09-11 9:39 Johannes Luber
[not found] ` <m3zl91zsqd.fsf@neno.mitica>
0 siblings, 1 reply; 2+ messages in thread
From: Johannes Luber @ 2009-09-11 9:39 UTC (permalink / raw)
To: qemu-devel
Hi,
I've been looking through the sources and stumbled over targphys.h, which contains:
/* target_phys_addr_t is the type of a physical address (its size can
be different from 'target_ulong'). We have sizeof(target_phys_addr)
= max(sizeof(unsigned long),
sizeof(size_of_target_physical_address)) because we must pass a
host pointer to memory operations in some cases */
#if TARGET_PHYS_ADDR_BITS == 32
typedef uint32_t target_phys_addr_t;
#define TARGET_PHYS_ADDR_MAX UINT32_MAX
#define TARGET_FMT_plx "%08x"
#elif TARGET_PHYS_ADDR_BITS == 64
typedef uint64_t target_phys_addr_t;
#define TARGET_PHYS_ADDR_MAX UINT64_MAX
#define TARGET_FMT_plx "%016" PRIx64
#endif
The comment says "sizeof(target_phys_addr) = max(...)" yet I don't the actual definition depending on the host at all. This means that on 64 bit hosts things are going wrong. sizeof(uint32_t) will be always 4 and actually using sizeof(unsigned long) will still result on 64-bit Windows in 4, despite pointer having 8 bytes.
As the comment remarks this won't work with physical pointers which reference pages beyond the 4 GB limit (unless the pointers don't have always some high bits set). So my question is: Does Qemu support 64-bit compilation for hosts correctly?
Best regards,
Johannes
--
Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate
für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02
^ permalink raw reply [flat|nested] 2+ messages in thread
* [Qemu-devel] Re: target_phys_addr_t definition
[not found] ` <m3zl91zsqd.fsf@neno.mitica>
@ 2009-09-11 11:05 ` Johannes Luber
0 siblings, 0 replies; 2+ messages in thread
From: Johannes Luber @ 2009-09-11 11:05 UTC (permalink / raw)
To: Juan Quintela; +Cc: qemu-devel
> "Johannes Luber" <JALuber@gmx.de> wrote:
...
> So my question is: Does Qemu support 64-bit compilation
> for hosts correctly?
> >
> > Best regards,
> > Johannes
>
> Dragons here.
>
> See configure, the interesting bit is:
>
> if [ $target_phys_bits -lt $hostlongbits ] ; then
> target_phys_bits=$hostlongbits
> fi
>
> TARGET_PHYS_ADDR_BITS is the maximum of host + target long sizes.
>
> Later, Juan.
Thanks for the quick response! It seems that someone edited my local config.h which resulted in losing the config-host.h. It would be nice if something like
#ifndef HOST_LONG_BITS
#error No HOST_LONG_BITS
#endif
could be included at appropriate places. This would make this problem self-evident in the future.
Best regards,
Johannes
--
Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3 -
sicherer, schneller und einfacher! http://portal.gmx.net/de/go/atbrowser
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2009-09-11 11:05 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-09-11 9:39 [Qemu-devel] target_phys_addr_t definition Johannes Luber
[not found] ` <m3zl91zsqd.fsf@neno.mitica>
2009-09-11 11:05 ` [Qemu-devel] " Johannes Luber
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).