This is the mail archive of the
mailing list for the binutils project.
Re: [PATCH] [RFC] [GOLD] s390 -fsplit-stack support.
- From: Cary Coutant <ccoutant at gmail dot com>
- To: Marcin KoÅcielnicki <koriakin at 0x04 dot net>
- Cc: Binutils <binutils at sourceware dot org>
- Date: Sun, 13 Dec 2015 14:42:23 -0800
- Subject: Re: [PATCH] [RFC] [GOLD] s390 -fsplit-stack support.
- Authentication-results: sourceware.org; auth=none
- References: <1449674134-27969-1-git-send-email-koriakin at 0x04 dot net> <CAJimCsEE+wNm-jMKAMGbRynY6sVNDTNXhtjvHz=QyB1tmqa3fA at mail dot gmail dot com> <CAJimCsFPTWkw98DW_RMYaJixkTTiCT4YoOH0oteCfhp=9DPDXQ at mail dot gmail dot com> <CAJimCsHZ877+ExvSKVd5p-x4CZHErLXT=rY_6OaWA_1T9P17Cg at mail dot gmail dot com> <566CD349 dot 5050202 at 0x04 dot net>
> Thanks for the patches! I've updated mine, and the changes are now limitted
> to s390-specific code. The output_view patch has a minor problem: it
> returns a const unsigned char *, but I need a read-write view (I have to
> bump the frame size in the parameter block). I've just const_casted around
> it for now, but I suppose we should change the return type instead?
Oops, sorry! I've committed the attached patch to remove the const.
2015-12-13 Cary Coutant <firstname.lastname@example.org>
* object.h (Object::get_output_view): remove const from return type.
* reloc.cc (Sized_relobj_file::do_get_output_view): Likewise.
> Also, I still have an icky static_cast in the "find parameter block"
> sequence. but I suppose it's here to stay:
> + Sized_relobj_file<size, true> *object_sized =
> + static_cast<Sized_relobj_file<size, true> *>(object);
Yes, we have a few other similar casts. It's not a high priority, but
I'll eventually see about adding a virtual function to the Object
class to return a Sized_relobj_file pointer to itself.
> Btw, this patch shouldn't be landing yet, I still have to finish the gcc
No problem; let me know when it's ready.