[patch] original patch from mailing list

sparc-zero.got.patch (text/plain), 6.17 KB, created by Rolf Eike Beer on 2018-04-02 16:46:02 UTC

From [e-mail] Fri Dec 08 01:19:38 2017
Return-path: <[e-mail]>
Envelope-to: [e-mail]
Received: by mailer.emlix.com
	id 1eN7Lj-0001ix-6o; Fri, 08 Dec 2017 02:21:11 +0100
Received: from mx1.emlix.com ([46.4.235.150]:41002)
	by gate.emlix.com with esmtps (TLSv1.2:DHE-RSA-AES256-GCM-SHA384:256)
	(Exim 4.82_1-5b7a7c0-XX)
	(envelope-from <[e-mail]>)
	id 1eN7KL-0007e3-2b
	for [e-mail]; Fri, 08 Dec 2017 02:19:46 +0100
X-CTCH-RefID: str=0001.0A090204.5A29E8B2.003B,ss=1,re=0.000,recu=0.000,reip=0.000,cl=1,cld=1,fgs=0
Received-SPF: pass (gmail.com ... _spf.google.com: Sender is authorized to use '[e-mail]' in 'mfrom' identity (mechanism 'include:_netblocks.google.com' matched)) receiver=mx1.emlix.com; identity=mailfrom; envelope-from="[e-mail]"; helo=mail-pf0-f176.google.com; client-ip=209.85.192.176
Received: from mail-pf0-f176.google.com (mail-pf0-f176.google.com [209.85.192.176])
	(using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits))
	(No client certificate requested)
	by mx1.emlix.com (Postfix) with ESMTPS id 4CF9825F21C
	for <[e-mail]>; Fri,  8 Dec 2017 02:19:45 +0100 (CET)
Received: by mail-pf0-f176.google.com with SMTP id u25so6043493pfg.5
        for <[e-mail]>; Thu, 07 Dec 2017 17:19:44 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=gmail.com; s=20161025;
        h=date:from:to:cc:subject:message-id:references:mime-version
         :content-disposition:in-reply-to:user-agent;
        bh=PBSyJWqX0xWOvHCkBs8Ihevl7ryqoqCycatAt5mY+jk=;
        b=GDR7/YXBZm00Eo5Pv2IskXG7LWCu+4MU82lOKbKFXHD08Vjm6QVCsjNXQV7Rv/o8se
         zzr5bqlWYZzTxD0MI/FuDSc4IhW+Xbn40nJb9MRdMrCUbTngdxp93tCLMxjzyD7y1tt0
         9pytMOzqBvVJQ+Bi1WNJVOAOxUeUeUDXTIcVZ4pyR+CWNqG0oQOLnALto9OMyoaYB2Ib
         7mlIoj/t/somzc02WVjeWwxYIfd8dioIch9rV2AqYezVumtCte+TNVWMqt5Ohv5Mx5I5
         krx0O6odZurKyrqr2Rk9YkEK7wSRlwqeVUQ8Im94pgd7XzJwoxmCjbvWLjv8iEIs+75V
         TarQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=1e100.net; s=20161025;
        h=x-gm-message-state:date:from:to:cc:subject:message-id:references
         :mime-version:content-disposition:in-reply-to:user-agent;
        bh=PBSyJWqX0xWOvHCkBs8Ihevl7ryqoqCycatAt5mY+jk=;
        b=A6hINwfPWNtRauIWEfo64BqudyKEGBbaV+e5tKOrFCkHT1sBXQodfJTdKTmgZP3jmr
         YjrdFZAUv/DaP8Gtbj6XIU4/lYU7FJc3v0VnxYPX0AxS1G73e0qeYktvAE0EEuH9vSe1
         jG2/faXoGDtd7MZjBBmlU8oMkJbFLBcAAhpzTiaKmPj7ibKcOUfmHvoDs9H/RVFBsBLV
         ZsPg1dg9l9emQ/WZKae0ffP2sNtBrwUhD1ELi6gjA/UBBqDRWAlMoJGe5jxE4c1UypDV
         dCThJbmTZnGqSylZrCwjCgvyMyNLiZoxl+mD9GzAikicQhTJECwCMPjlAhn1WXRU/B7b
         k+3g==
X-Gm-Message-State: AJaThX5zx9Zzlxus4m1HCIGz7NApjxSMfI13Hv3bci/ke7AgMFoQX2BM
	TJgqtSCcXuD6yqWLu11iZrbWvA==
X-Google-Smtp-Source: AGs4zMbFKvW5hzulZ42+gUcdCrHr4uLDOw0rII+JYwbcNyMX2ZZNlbN3Cvlzo1G9DPB17ENvrGXT9A==
X-Received: by 10.84.240.193 with SMTP id l1mr28243752plt.240.1512695982957;
        Thu, 07 Dec 2017 17:19:42 -0800 (PST)
Received: from bubble.grove.modra.org (CPE-58-175-244-173.hdcz1.win.bigpond.net.au. [58.175.244.173])
        by smtp.gmail.com with ESMTPSA id v43sm10787378pgn.65.2017.12.07.17.19.41
        (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128);
        Thu, 07 Dec 2017 17:19:41 -0800 (PST)
Received: by bubble.grove.modra.org (Postfix, from userid 1000)
	id 14DB1C0729; Fri,  8 Dec 2017 11:49:38 +1030 (ACDT)
Date: Fri, 8 Dec 2017 11:49:38 +1030
From: Alan Modra <[e-mail]>
To: Rolf Eike Beer <[e-mail]>
Cc: [e-mail]
Subject: Re: Binutils 2.29.1 breaks libpcap on sparc32 by filling .got
Message-ID: <[e-mail]>
References: <13757039.pIICRVicT9@devpool21>
 <[e-mail]>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <[e-mail]>
User-Agent: Mutt/1.5.24 (2015-08-30)

On Fri, Dec 08, 2017 at 11:20:12AM +1030, Alan Modra wrote:
> So it looks like sparc and sparc64 ld needs to be sure to zero section
> contents whenever a RELATIVE relocation is emitted, to avoid this
> glibc bug.

Please test this.

	* elfxx-sparc.c (_bfd_sparc_elf_relocate_section): When emitting
	dynamic R_SPARC_RELATIVE for GOT entries, ensure the section
	contents are zeroed.

diff --git a/bfd/elfxx-sparc.c b/bfd/elfxx-sparc.c
index adc9ed3..9885059 100644
--- a/bfd/elfxx-sparc.c
+++ b/bfd/elfxx-sparc.c
@@ -3212,10 +3212,6 @@ _bfd_sparc_elf_relocate_section (bfd *output_bfd,
 		    off &= ~1;
 		  else
 		    {
-		      SPARC_ELF_PUT_WORD (htab, output_bfd, relocation,
-					  htab->elf.sgot->contents + off);
-		      h->got.offset |= 1;
-
 		      if (h->dynindx == -1
 			  && !h->forced_local
 			  && h->root.type != bfd_link_hash_undefweak
@@ -3225,6 +3221,10 @@ _bfd_sparc_elf_relocate_section (bfd *output_bfd,
 			     generate R_SPARC_RELATIVE here.  */
 			  relative_reloc = TRUE;
 			}
+		      else
+			SPARC_ELF_PUT_WORD (htab, output_bfd, relocation,
+					    htab->elf.sgot->contents + off);
+		      h->got.offset |= 1;
 		    }
 		}
 	      else
@@ -3245,12 +3245,10 @@ _bfd_sparc_elf_relocate_section (bfd *output_bfd,
 	      else
 		{
 		  if (bfd_link_pic (info))
-		    {
-		      relative_reloc = TRUE;
-		    }
-
-		  SPARC_ELF_PUT_WORD (htab, output_bfd, relocation,
-				      htab->elf.sgot->contents + off);
+		    relative_reloc = TRUE;
+		  else
+		    SPARC_ELF_PUT_WORD (htab, output_bfd, relocation,
+					htab->elf.sgot->contents + off);
 		  local_got_offsets[r_symndx] |= 1;
 		}
 	    }
@@ -3271,8 +3269,14 @@ _bfd_sparc_elf_relocate_section (bfd *output_bfd,
 	      outrel.r_info = SPARC_ELF_R_INFO (htab, NULL,
 						0, R_SPARC_RELATIVE);
 	      outrel.r_addend = relocation;
-	      relocation = 0;
 	      sparc_elf_append_rela (output_bfd, s, &outrel);
+	      /* Versions of glibc ld.so at least up to 2.26 wrongly
+		 add the section contents to the value calculated for
+		 a RELATIVE reloc.  Zero the contents to work around
+		 this bug.  */
+	      relocation = 0;
+	      SPARC_ELF_PUT_WORD (htab, output_bfd, relocation,
+				  htab->elf.sgot->contents + off);
 	    }
 
 	  relocation = htab->elf.sgot->output_offset + off - got_base;

-- 
Alan Modra
Australia Development Lab, IBM