public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
From: liuzhensong <liuzhensong@loongson.cn>
To: Xi Ruoyao <xry111@xry111.site>, binutils@sourceware.org
Cc: Lulu Cheng <chenglulu@loongson.cn>, Wang Xuerui <i@xen0n.name>,
	Chenghua Xu <xuchenghua@loongson.cn>
Subject: Re: [PATCH 2/2] LoongArch: Fix R_LARCH_IRELATIVE insertion after elf_link_sort_relocs
Date: Thu, 15 Sep 2022 21:03:33 +0800	[thread overview]
Message-ID: <939eebd1-365d-72c5-5b2c-43bd6e0f1ff2@loongson.cn> (raw)
In-Reply-To: <20220913154414.554861-3-xry111@xry111.site>


在 2022/9/13 下午11:44, Xi Ruoyao 写道:
> loongarch_elf_finish_dynamic_symbol is called after elf_link_sort_relocs
> if -z combreloc.  elf_link_sort_relocs redistributes the contents of
> .rela.* sections those would be merged into .rela.dyn, so the slot for
> R_LARCH_IRELATIVE may be out of relplt->contents now.

It May be unnecessary find the slot in section of .rela.dyn.

The slot of R_LARCH_IRELATIVE is within the scope relplt->contents (the 
space

is calculated in function local_allocate_ifunc_dyn_relocs).

We can find the slot just in relplt->contents(Only for dynamic ifunc. I 
did not test for static ifunc.).

>
> To make things worse, the boundary check
>
>      dyn < dyn + relplt->size / sizeof (*dyn)
>
> is obviously wrong ("x + 10 < x"? :), causing the issue undetected
> during the linking process and the resulted executable suddenly crashes
> at runtime.
>
> The issue was found during an attempt to add static-pie support to the
> toolchain.
>
> Fix it by iterating through the inputs of .rela.dyn to find the slot.
> ---
>   bfd/elfnn-loongarch.c | 41 ++++++++++++++++++++++++-----------------
>   1 file changed, 24 insertions(+), 17 deletions(-)
>
> diff --git a/bfd/elfnn-loongarch.c b/bfd/elfnn-loongarch.c
> index 4b408b1db72..1de9dbfecfa 100644
> --- a/bfd/elfnn-loongarch.c
> +++ b/bfd/elfnn-loongarch.c
> @@ -3511,6 +3511,12 @@ loongarch_elf_finish_dynamic_symbol (bfd *output_bfd,
>   {
>     struct loongarch_elf_link_hash_table *htab = loongarch_elf_hash_table (info);
>     const struct elf_backend_data *bed = get_elf_backend_data (output_bfd);
> +  asection *rela_dyn = bfd_get_section_by_name (output_bfd, ".rela.dyn");
> +  struct bfd_link_order *lo = NULL;
> +  Elf_Internal_Rela *slot = NULL, *last_slot = NULL;
> +
> +  if (rela_dyn)
> +    lo = rela_dyn->map_head.link_order;
>   
>     if (h->plt.offset != MINUS_ONE)
>       {
> @@ -3520,6 +3526,7 @@ loongarch_elf_finish_dynamic_symbol (bfd *output_bfd,
>         uint32_t plt_entry[PLT_ENTRY_INSNS];
>         bfd_byte *loc;
>         Elf_Internal_Rela rela;
> +      asection *rela_sec = NULL;
>   
>         if (htab->elf.splt)
>   	{
> @@ -3572,31 +3579,31 @@ loongarch_elf_finish_dynamic_symbol (bfd *output_bfd,
>   	  && (relplt == htab->elf.srelgot
>   	      || relplt == htab->elf.irelplt))
>   	{
> -	    {
> -	      rela.r_info = ELFNN_R_INFO (0, R_LARCH_IRELATIVE);
> -	      rela.r_addend = (h->root.u.def.value
> +	  rela.r_info = ELFNN_R_INFO (0, R_LARCH_IRELATIVE);
> +	  rela.r_addend = (h->root.u.def.value
>   			       + h->root.u.def.section->output_section->vma
>   			       + h->root.u.def.section->output_offset);
> -	    }
>   
> -	    /* Find the space after dyn sort.  */
> +	  /* Find the space after dyn sort.  */
> +	  while (slot == last_slot || slot->r_offset != 0)
>   	    {
> -	      Elf_Internal_Rela *dyn = (Elf_Internal_Rela *)relplt->contents;
> -	      bool fill = false;
> -	      for (;dyn < dyn + relplt->size / sizeof (*dyn); dyn++)
My mistake.
> +	      if (slot != last_slot)
>   		{
> -		  if (0 == dyn->r_offset)
> -		    {
> -		      bed->s->swap_reloca_out (output_bfd, &rela,
> -					       (bfd_byte *)dyn);
> -		      relplt->reloc_count++;
> -		      fill = true;
> -		      break;
> -		    }
> +		  slot++;
> +		  continue;
>   		}
> -	      BFD_ASSERT (fill);
> +
> +	      BFD_ASSERT (lo != NULL);
> +	      rela_sec = lo->u.indirect.section;
> +	      lo = lo->next;
> +
> +	      slot = (Elf_Internal_Rela *)rela_sec->contents;
> +	      last_slot = (Elf_Internal_Rela *)(rela_sec->contents +
> +						rela_sec->size);
>   	    }
>   
> +	  bed->s->swap_reloca_out (output_bfd, &rela, (bfd_byte *)slot);
> +	  rela_sec->reloc_count++;
>   	}
>         else
>   	{


  reply	other threads:[~2022-09-15 13:03 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-13 15:44 [PATCH 0/2] LoongArch: Fix two bugs breaking IFUNC in static PIE Xi Ruoyao
2022-09-13 15:44 ` [PATCH 1/2] LoongArch: Avoid heap-buffer-overflow in loongarch_elf_relocate_section Xi Ruoyao
2022-09-14  8:57   ` liuzhensong
2022-09-14 10:15     ` Xi Ruoyao
2022-09-14 11:15       ` Xi Ruoyao
2022-09-15  1:47         ` liuzhensong
2022-09-15  2:56           ` Xi Ruoyao
2022-09-15  3:54             ` liuzhensong
2022-09-15  6:13               ` Xi Ruoyao
2022-09-13 15:44 ` [PATCH 2/2] LoongArch: Fix R_LARCH_IRELATIVE insertion after elf_link_sort_relocs Xi Ruoyao
2022-09-15 13:03   ` liuzhensong [this message]
2022-09-16 20:11     ` Xi Ruoyao
2022-09-18  9:58       ` Xi Ruoyao

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=939eebd1-365d-72c5-5b2c-43bd6e0f1ff2@loongson.cn \
    --to=liuzhensong@loongson.cn \
    --cc=binutils@sourceware.org \
    --cc=chenglulu@loongson.cn \
    --cc=i@xen0n.name \
    --cc=xry111@xry111.site \
    --cc=xuchenghua@loongson.cn \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).