1030400 - Add `JS::Symbol *sym;` field to jsval_layout, ifdef JS_NUNBOX32, if !defined(IS_LITTLE_ENDIAN)

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jason Orendorff [:jorendorff]]

Spotted by jfkthame (see bug 645416 comment 166).

I guess no one is building SpiderMonkey with JS_NUNBOX32 on any big-endian platforms. I've never tried to build it for such a platform. Do they exist? Maybe this isn't even the only thing broken there.

Still, it's easy enough to add the field, since it's obviously busted otherwise.

Comment 1

[Nathan Froyd [:froydnj]]

(In reply to Jason Orendorff [:jorendorff] from comment #0)
> Spotted by jfkthame (see bug 645416 comment 166).
> 
> I guess no one is building SpiderMonkey with JS_NUNBOX32 on any big-endian
> platforms. I've never tried to build it for such a platform. Do they exist?
> Maybe this isn't even the only thing broken there.

Sparc, PPC, and MIPS, according to configure.in.  I know Landry builds for sparc (though maybe only 64-bit, which is PUNBOX64) and occasionally PPC.

Comment 2

[Steve Singer (:stevensn)]

Attachment: bug1030400.diff

I regularly build firefox on ppc32.
The attached patch adds sym to the conditional branch

Comment 3

[Jason Orendorff [:jorendorff]]

Comment on attachment 8446286 [details] [diff] [review]
bug1030400.diff

Review of attachment 8446286 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks. Sorry for the oversight.

Comment 4

[Kartikaya Gupta (email:[e-mail])]

(In reply to Jason Orendorff [:jorendorff] from comment #0)
> Spotted by jfkthame (see bug 645416 comment 166).
> 
> I guess no one is building SpiderMonkey with JS_NUNBOX32 on any big-endian
> platforms. I've never tried to build it for such a platform. Do they exist?
> Maybe this isn't even the only thing broken there.

Maybe B2G for hamachi is such a platform. I ran into a build failure due to missing "sym". I'll try this patch and see if it fixes it.

Comment 5

[Kartikaya Gupta (email:[e-mail])]

This patch seems to have fixed my build issue. For the record, I was building master B2G for the hamachi device on an OS X machine. I assume tbpl would have gone red if the hamachi builds had been broken across the board. From the code it doesn't look like this is compiled for the host so I'm not sure why it would fail for me on OS X but pass on the buildbot machines. And yes, I did try a clobber build (rm -rf objdir-gecko) without this patch and it was still failing. A similar clobber build with the patch succeeded.

Comment 6

[Steve Singer (:stevensn)]

https://tbpl.mozilla.org/?tree=Try&rev=bfd6e52448cc
Builds fine on a tier-1 32 bit platforms (the patch shouldn't effect them)

Comment 7

[Steve Singer (:stevensn)]

Kartikaya, 
What cpu/platform is the hamachi?

Comment 8

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/433d3d69cecb

Comment 9

[Kartikaya Gupta (email:[e-mail])]

Hamachi has a Qualcomm MSM7227A chipset.

Comment 10

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/433d3d69cecb