RE: [PATCH] MIPS support for --hash-style=gnu

["Neil Schellenberger (neschell)" <neschell@cisco.com>]

[Apologies for the long delay: I have been out of the office....]

Hmm, good point.

In a separate email chain, it was also noted that DT_MIPS_SYMTABNO might be used to compute ngnusyms.
I originally avoided DT_MIPS_SYMTABNO only because I wasn't absolutely certain that it was guaranteed 
that it (the number of entries in the .dynsym section) was always going to be the same as the the number of 
entries in chains (plus symidx) so I decided to play it safe.  That may well be needless paranoia on my part.

Opinions?

Regards,
Neil

> -----Original Message-----
> From: Koding [mailto:majiang31312@majiang31312]
> Sent: Tuesday, January 05, 2016 1:11 AM
> To: binutils@sourceware.org
> Subject: Re: [PATCH] MIPS support for --hash-style=gnu
> 
> Hi all,
> 	I am using this patch for a loongson 3A cpu。After some tests , I
> found that there was a mistake that cause many many alignment-faults in
> this patch.
> 	The patch added a Elf32_Word ngnusyms in the head of .gnu.xhash
> section, then "ElfW(Addr)  bitmask[maskwords]"  became unaligned(8-
> bytes).
> 	IMO, we should  change "Elf32_Word ngnusyms" to "Elf32_Xword
> ngnusyms" , or add 4-bytes pad after ngnusyms.