LGTM I think the shift might have been equivalent except that it would be wrong ...
13 years, 5 months ago
(2010-11-15 21:45:56 UTC)
#2
LGTM
I think the shift might have been
equivalent except that it would be
wrong for 64-bit values because int is 32 bits.
Please change (uchar *) to (uchar*).
On 2010/11/15 21:45:56, rsc wrote: > LGTM > > I think the shift might have ...
13 years, 5 months ago
(2010-11-16 04:23:19 UTC)
#5
On 2010/11/15 21:45:56, rsc wrote:
> LGTM
>
> I think the shift might have been
> equivalent except that it would be
> wrong for 64-bit values because int is 32 bits.
Shifting with an uvlong cast works on a little-ending host but not on a
big-endian one.
Unfortunately, inuxi8 is still wrong when building on a host that has the
opposite byte order of the target. That bug comes directly from the inferno
tools.
> Unfortunately, inuxi8 is still wrong when building on a host that has > the ...
13 years, 5 months ago
(2010-11-16 12:55:46 UTC)
#7
> Unfortunately, inuxi8 is still wrong when building on a host that has
> the opposite byte order of the target.
Why is that? Why is it different from inuxi4?
Russ
On 16 November 2010 04:55, Russ Cox <rsc@golang.org> wrote: >> Unfortunately, inuxi8 is still wrong ...
13 years, 5 months ago
(2010-11-16 15:04:43 UTC)
#8
On 16 November 2010 04:55, Russ Cox <rsc@golang.org> wrote:
>> Unfortunately, inuxi8 is still wrong when building on a host that has
>> the opposite byte order of the target.
>
> Why is that? Why is it different from inuxi4?
Inuxi8 is initialised as the catenation of inuxi4 and inuxi4+4. So on
a little-endian host we get,
inuxi4=0123
inuxi8=01234567
which is right, but on a big-endian host it's,
inuxi4=3210
inuxi8=32107654
As far as I can tell inuxi8 should be 76543210 on a big-endian.
About to do a PTAL with it fixed. It's ugly though.
Going in a different direction, maybe the nuxi stuff should be
replaced by the corresponding get and put byte packing functions. Nuxi
is only used in a couple of places (and only in ld).
Eoghan
Issue 3002041: code review 3002041: ld: fix dwarf decoding of 64-bit reflect values
(Closed)
Created 13 years, 5 months ago by ejsherry
Modified 13 years, 4 months ago
Reviewers:
Base URL:
Comments: 0