[RFC PATCH 3/3] add r_debug multiple namespaces support

[RFC PATCH 3/3] add r_debug multiple namespaces support

[Daniel Walker via Libc-alpha]

From: Conan C Huang <conhuang@cisco.com>

GLIBC Bugzilla: 15971
GDB Bugzilla: 11839

Glibc does not provide an interface for debugger to access libraries loaded in
multiple namespaces via dlmopen.

The current rtld-debugger interface is described in the file
elf/rtld-debugger-interface.txt under the "Standard debugger interface" heading.
This interface only provides access to the first link-map (LM_ID_BASE).

This solution converts r_debug into a linked-list, where each element correlates
to an unique namespace. Debugger (GDB) can traverse r_debug linked-list to
retrieve information for all loaded namespaces.
---
 elf/dl-debug.c | 13 ++++++++++---
 elf/link.h     |  4 ++++
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/elf/dl-debug.c b/elf/dl-debug.c
index 4b3d3ad6ba..7eba477fd3 100644
--- a/elf/dl-debug.c
+++ b/elf/dl-debug.c
@@ -44,20 +44,27 @@ struct r_debug _r_debug;
 struct r_debug *
 _dl_debug_initialize (ElfW(Addr) ldbase, Lmid_t ns)
 {
-  struct r_debug *r;
+  struct r_debug *r, *rp;
 
   if (ns == LM_ID_BASE)
     r = &_r_debug;
   else
-    r = &GL(dl_ns)[ns]._ns_debug;
+    {
+      r = &GL(dl_ns)[ns]._ns_debug;
+      rp = &GL(dl_ns)[ns - 1]._ns_debug;
+      rp->next = r;
+      if (ns - 1 == LM_ID_BASE)
+        _r_debug.next = r;
+    }
 
   if (r->r_map == NULL || ldbase != 0)
     {
       /* Tell the debugger where to find the map of loaded objects.  */
-      r->r_version = 1	/* R_DEBUG_VERSION XXX */;
+      r->r_version = 3  /* R_DEBUG_VERSION XXX */;
       r->r_ldbase = ldbase ?: _r_debug.r_ldbase;
       r->r_map = (void *) GL(dl_ns)[ns]._ns_loaded;
       r->r_brk = (ElfW(Addr)) &_dl_debug_state;
+      r->next = NULL;
     }
 
   return r;
diff --git a/elf/link.h b/elf/link.h
index 0048ad5d4d..5a42511636 100644
--- a/elf/link.h
+++ b/elf/link.h
@@ -61,6 +61,10 @@ struct r_debug
       } r_state;
 
     ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
+
+    /* Link to next r_debug struct. Each r_debug struct represents a
+       different namespace.  The first r_debug struct is the default.  */
+    struct r_debug *next;
   };
 
 /* This is the instance of that structure used by the dynamic linker.  */
-- 
2.17.1

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Florian Weimer via Libc-alpha]

* Daniel Walker via Libc-alpha:

> diff --git a/elf/link.h b/elf/link.h
> index 0048ad5d4d..5a42511636 100644
> --- a/elf/link.h
> +++ b/elf/link.h
> @@ -61,6 +61,10 @@ struct r_debug
>        } r_state;
>  
>      ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
> +
> +    /* Link to next r_debug struct. Each r_debug struct represents a
> +       different namespace.  The first r_debug struct is the default.  */
> +    struct r_debug *next;
>    };
>  
>  /* This is the instance of that structure used by the dynamic linker.  */

How has this patch been tested?  I expect that it will cause an abilist
mismatch for the _r_debug symbol in the dynamic linker.

If we go this route to add this capability, I think we have to add a new
symbol version for the _r_debug symbol, and keep the old one at the
previous size.

How is your compatibility experience with the size and version change?
How many tools need updating before they work again?

A different approach would add another symbol (parallel to _r_debug) to
store this data.  This would avoid the need for any immediate tool
updates.

Thanks,
Florian

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Carlos O'Donell via Libc-alpha]

On 6/26/20 5:05 PM, Florian Weimer via Libc-alpha wrote:
> * Daniel Walker via Libc-alpha:
> 
>> diff --git a/elf/link.h b/elf/link.h
>> index 0048ad5d4d..5a42511636 100644
>> --- a/elf/link.h
>> +++ b/elf/link.h
>> @@ -61,6 +61,10 @@ struct r_debug
>>        } r_state;
>>  
>>      ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
>> +
>> +    /* Link to next r_debug struct. Each r_debug struct represents a
>> +       different namespace.  The first r_debug struct is the default.  */
>> +    struct r_debug *next;
>>    };
>>  
>>  /* This is the instance of that structure used by the dynamic linker.  */
> 
> How has this patch been tested?  I expect that it will cause an abilist
> mismatch for the _r_debug symbol in the dynamic linker.
> 
> If we go this route to add this capability, I think we have to add a new
> symbol version for the _r_debug symbol, and keep the old one at the
> previous size.
> 
> How is your compatibility experience with the size and version change?
> How many tools need updating before they work again?
> 
> A different approach would add another symbol (parallel to _r_debug) to
> store this data.  This would avoid the need for any immediate tool
> updates.

I mention this in my response to the cover letter in this series.

This patch is probably unacceptable as-is because of application
expectations.

-- 
Cheers,
Carlos.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Florian Weimer via Libc-alpha]

* Carlos O'Donell:

> On 6/26/20 5:05 PM, Florian Weimer via Libc-alpha wrote:
>> * Daniel Walker via Libc-alpha:
>> 
>>> diff --git a/elf/link.h b/elf/link.h
>>> index 0048ad5d4d..5a42511636 100644
>>> --- a/elf/link.h
>>> +++ b/elf/link.h
>>> @@ -61,6 +61,10 @@ struct r_debug
>>>        } r_state;
>>>  
>>>      ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
>>> +
>>> +    /* Link to next r_debug struct. Each r_debug struct represents a
>>> +       different namespace.  The first r_debug struct is the default.  */
>>> +    struct r_debug *next;
>>>    };
>>>  
>>>  /* This is the instance of that structure used by the dynamic linker.  */
>> 
>> How has this patch been tested?  I expect that it will cause an abilist
>> mismatch for the _r_debug symbol in the dynamic linker.
>> 
>> If we go this route to add this capability, I think we have to add a new
>> symbol version for the _r_debug symbol, and keep the old one at the
>> previous size.
>> 
>> How is your compatibility experience with the size and version change?
>> How many tools need updating before they work again?
>> 
>> A different approach would add another symbol (parallel to _r_debug) to
>> store this data.  This would avoid the need for any immediate tool
>> updates.
>
> I mention this in my response to the cover letter in this series.

Your explanation there was truncated.

> This patch is probably unacceptable as-is because of application
> expectations.

But perhaps Cisco's experience shows that our worries are unfounded?

Thanks,
Florian

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Carlos O'Donell via Libc-alpha]

On 6/26/20 5:24 PM, Florian Weimer wrote:
> * Carlos O'Donell:
> 
>> On 6/26/20 5:05 PM, Florian Weimer via Libc-alpha wrote:
>>> * Daniel Walker via Libc-alpha:
>>>
>>>> diff --git a/elf/link.h b/elf/link.h
>>>> index 0048ad5d4d..5a42511636 100644
>>>> --- a/elf/link.h
>>>> +++ b/elf/link.h
>>>> @@ -61,6 +61,10 @@ struct r_debug
>>>>        } r_state;
>>>>  
>>>>      ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
>>>> +
>>>> +    /* Link to next r_debug struct. Each r_debug struct represents a
>>>> +       different namespace.  The first r_debug struct is the default.  */
>>>> +    struct r_debug *next;
>>>>    };
>>>>  
>>>>  /* This is the instance of that structure used by the dynamic linker.  */
>>>
>>> How has this patch been tested?  I expect that it will cause an abilist
>>> mismatch for the _r_debug symbol in the dynamic linker.
>>>
>>> If we go this route to add this capability, I think we have to add a new
>>> symbol version for the _r_debug symbol, and keep the old one at the
>>> previous size.
>>>
>>> How is your compatibility experience with the size and version change?
>>> How many tools need updating before they work again?
>>>
>>> A different approach would add another symbol (parallel to _r_debug) to
>>> store this data.  This would avoid the need for any immediate tool
>>> updates.
>>
>> I mention this in my response to the cover letter in this series.
> 
> Your explanation there was truncated.

Truncated in which way?

>> This patch is probably unacceptable as-is because of application
>> expectations.
> 
> But perhaps Cisco's experience shows that our worries are unfounded?

That would be great!

I would want to see gdb changed to r_version > 0, and document that
the increasing r_version means changes that only expand the structure
and add new information while keeping old information backwards
compatible.

I'm not sure it would work to version _r_debug, since the debugger
is using DT_DEBUG and we only get to put one value in that
.dynamic entry.

-- 
Cheers,
Carlos.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Daniel Walker (danielwa) via Libc-alpha]

On Fri, Jun 26, 2020 at 11:24:21PM +0200, Florian Weimer wrote:
> * Carlos O'Donell:
> 
> > On 6/26/20 5:05 PM, Florian Weimer via Libc-alpha wrote:
> >> * Daniel Walker via Libc-alpha:
> >> 
> >>> diff --git a/elf/link.h b/elf/link.h
> >>> index 0048ad5d4d..5a42511636 100644
> >>> --- a/elf/link.h
> >>> +++ b/elf/link.h
> >>> @@ -61,6 +61,10 @@ struct r_debug
> >>>        } r_state;
> >>>  
> >>>      ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
> >>> +
> >>> +    /* Link to next r_debug struct. Each r_debug struct represents a
> >>> +       different namespace.  The first r_debug struct is the default.  */
> >>> +    struct r_debug *next;
> >>>    };
> >>>  
> >>>  /* This is the instance of that structure used by the dynamic linker.  */
> >> 
> >> How has this patch been tested?  I expect that it will cause an abilist
> >> mismatch for the _r_debug symbol in the dynamic linker.
> >> 
> >> If we go this route to add this capability, I think we have to add a new
> >> symbol version for the _r_debug symbol, and keep the old one at the
> >> previous size.
> >> 
> >> How is your compatibility experience with the size and version change?
> >> How many tools need updating before they work again?
> >> 
> >> A different approach would add another symbol (parallel to _r_debug) to
> >> store this data.  This would avoid the need for any immediate tool
> >> updates.
> >
> > I mention this in my response to the cover letter in this series.
> 
> Your explanation there was truncated.
> 
> > This patch is probably unacceptable as-is because of application
> > expectations.
> 
> But perhaps Cisco's experience shows that our worries are unfounded?

I don't know that we can confirm this. Per my understanding these changes have
been maintained as one off changes which haven't entered our product yet. We're
now currently trying to add this to our product. We use OpenEmbedded to build
an SDK which is then used in our product. The tooling in typical rebuilt
against any new glibc which we have, and we don't reuse binaries from the prior
builds which may have a different glibc (even different patch level).

Daniel

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Florian Weimer]

* Carlos O'Donell via Libc-alpha:

> Truncated in which way?

This part:

| Your proposed solution of bumping the version is unacceptable,
| and was last rejected by Roland McGrath. The problem is that
| when you bump the version the current 

> I'm not sure it would work to version _r_debug, since the debugger
> is using DT_DEBUG and we only get to put one value in that
> .dynamic entry.

The symbol version is needed to avoid problems due to copy relocations
if the symbol is referenced directly from the main program.  Without
that, the object could be truncated.  It's not a debugger
compatibility feature.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Carlos O'Donell via Libc-alpha]

On 6/27/20 5:34 AM, Florian Weimer wrote:
> * Carlos O'Donell via Libc-alpha:
> 
>> Truncated in which way?
> 
> This part:
> 
> | Your proposed solution of bumping the version is unacceptable,
> | and was last rejected by Roland McGrath. The problem is that
> | when you bump the version the current 

Thanks. "It's Friday" is my only excuse.

I did provide some of the original links to the discussion.

Roland, as a steward at the time, was worried about exactly what
we see in gdb, which is "r_version != 1" may have made it into tooling.

We can test this. We can try to deploy a similar solution in Fedora Rawhide
and declare the semantics as we expect them to be.

That is to say that r_version == 1, is the entire structure as we have it,
and r_version == 2 *adds* but does not remove from the structure. Since
the data is maintained by the implementation and the caller is only
inspecting the data it should work.

>> I'm not sure it would work to version _r_debug, since the debugger
>> is using DT_DEBUG and we only get to put one value in that
>> .dynamic entry.
> 
> The symbol version is needed to avoid problems due to copy relocations
> if the symbol is referenced directly from the main program.  Without
> that, the object could be truncated.  It's not a debugger
> compatibility feature.

Correct, but this violates *how* you're supposed to use _r_debug.

If you have a static executable you can get away with referencing
_r_debug directly, but in that case symbol versions don't matter, and
you have whatever version you have at the time.

In the dynamic case it is different. The symbol should be looked up
via DT_DEBUG only which always points to the library-local address
of the data object (and the most recent version). In effect this
bypasses the COPY relocation?

If an application uses _r_debug, the symbol, directly, then they
should get a static copy via the COPY relocation, and it will not
be updated after that. Perhaps we can arrange for such an initial
_r_debug to indicate it's not active or initialized?

IMO the library should use a local-only reference to the _r_debug to
avoid going through the global reference.

I'm not keen to admit that a COPY reloc of _r_debug should work.

-- 
Cheers,
Carlos.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Florian Weimer via Libc-alpha]

* Carlos O'Donell via Libc-alpha:

>>> I'm not sure it would work to version _r_debug, since the debugger
>>> is using DT_DEBUG and we only get to put one value in that
>>> .dynamic entry.
>> 
>> The symbol version is needed to avoid problems due to copy relocations
>> if the symbol is referenced directly from the main program.  Without
>> that, the object could be truncated.  It's not a debugger
>> compatibility feature.
>
> Correct, but this violates *how* you're supposed to use _r_debug.

If it is possible to link against it, we need to add the new symbol
version, in my opinion.

> In the dynamic case it is different. The symbol should be looked up
> via DT_DEBUG only which always points to the library-local address
> of the data object (and the most recent version). In effect this
> bypasses the COPY relocation?

How is this supposed to work if the dynamic linker does contain
DT_DEBUG?

I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
mapped at a random address even for ET_EXEC main programs, there must be
some other mechanism to locate it.

Thanks,
Florian

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[H.J. Lu via Libc-alpha]

On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> * Carlos O'Donell via Libc-alpha:
>
> >>> I'm not sure it would work to version _r_debug, since the debugger
> >>> is using DT_DEBUG and we only get to put one value in that
> >>> .dynamic entry.
> >>
> >> The symbol version is needed to avoid problems due to copy relocations
> >> if the symbol is referenced directly from the main program.  Without
> >> that, the object could be truncated.  It's not a debugger
> >> compatibility feature.
> >
> > Correct, but this violates *how* you're supposed to use _r_debug.
>
> If it is possible to link against it, we need to add the new symbol
> version, in my opinion.
>
> > In the dynamic case it is different. The symbol should be looked up
> > via DT_DEBUG only which always points to the library-local address
> > of the data object (and the most recent version). In effect this
> > bypasses the COPY relocation?
>
> How is this supposed to work if the dynamic linker does contain
> DT_DEBUG?
>
> I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> mapped at a random address even for ET_EXEC main programs, there must be
> some other mechanism to locate it.
>
> Thanks,
> Florian
>

I opened:

https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2

with a proposal to extend struct r_debug for libraries loaded with
dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
and GDB.

-- 
H.J.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Daniel Walker via Libc-alpha]

On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
> >
> > * Carlos O'Donell via Libc-alpha:
> >
> > >>> I'm not sure it would work to version _r_debug, since the debugger
> > >>> is using DT_DEBUG and we only get to put one value in that
> > >>> .dynamic entry.
> > >>
> > >> The symbol version is needed to avoid problems due to copy relocations
> > >> if the symbol is referenced directly from the main program.  Without
> > >> that, the object could be truncated.  It's not a debugger
> > >> compatibility feature.
> > >
> > > Correct, but this violates *how* you're supposed to use _r_debug.
> >
> > If it is possible to link against it, we need to add the new symbol
> > version, in my opinion.
> >
> > > In the dynamic case it is different. The symbol should be looked up
> > > via DT_DEBUG only which always points to the library-local address
> > > of the data object (and the most recent version). In effect this
> > > bypasses the COPY relocation?
> >
> > How is this supposed to work if the dynamic linker does contain
> > DT_DEBUG?
> >
> > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > mapped at a random address even for ET_EXEC main programs, there must be
> > some other mechanism to locate it.
> >
> > Thanks,
> > Florian
> >
> 
> I opened:
> 
> https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> 
> with a proposal to extend struct r_debug for libraries loaded with
> dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> and GDB.


I have an updated set of changes which add a new dynamic entry
DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
implementation. I'm open to support different implementations.

I found while working on this project that the clang linker adds 3 proprietary new
dynamic entries. So there is demand to make adding to the table easier.

Daniel

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[H.J. Lu via Libc-alpha]

On Tue, Jul 27, 2021 at 10:40 AM Daniel Walker <danielwa@cisco.com> wrote:
>
> On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> > On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> > <libc-alpha@sourceware.org> wrote:
> > >
> > > * Carlos O'Donell via Libc-alpha:
> > >
> > > >>> I'm not sure it would work to version _r_debug, since the debugger
> > > >>> is using DT_DEBUG and we only get to put one value in that
> > > >>> .dynamic entry.
> > > >>
> > > >> The symbol version is needed to avoid problems due to copy relocations
> > > >> if the symbol is referenced directly from the main program.  Without
> > > >> that, the object could be truncated.  It's not a debugger
> > > >> compatibility feature.
> > > >
> > > > Correct, but this violates *how* you're supposed to use _r_debug.
> > >
> > > If it is possible to link against it, we need to add the new symbol
> > > version, in my opinion.
> > >
> > > > In the dynamic case it is different. The symbol should be looked up
> > > > via DT_DEBUG only which always points to the library-local address
> > > > of the data object (and the most recent version). In effect this
> > > > bypasses the COPY relocation?
> > >
> > > How is this supposed to work if the dynamic linker does contain
> > > DT_DEBUG?
> > >
> > > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > > mapped at a random address even for ET_EXEC main programs, there must be
> > > some other mechanism to locate it.
> > >
> > > Thanks,
> > > Florian
> > >
> >
> > I opened:
> >
> > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> >
> > with a proposal to extend struct r_debug for libraries loaded with
> > dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> > and GDB.
>
>
> I have an updated set of changes which add a new dynamic entry
> DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
> implementation. I'm open to support different implementations.

Let's work on it after glibc 2.34 is branched.

> I found while working on this project that the clang linker adds 3 proprietary new
> dynamic entries. So there is demand to make adding to the table easier.
>
> Daniel



-- 
H.J.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[H.J. Lu via Libc-alpha]

On Tue, Jul 27, 2021 at 11:14 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Tue, Jul 27, 2021 at 10:40 AM Daniel Walker <danielwa@cisco.com> wrote:
> >
> > On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> > > On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> > > <libc-alpha@sourceware.org> wrote:
> > > >
> > > > * Carlos O'Donell via Libc-alpha:
> > > >
> > > > >>> I'm not sure it would work to version _r_debug, since the debugger
> > > > >>> is using DT_DEBUG and we only get to put one value in that
> > > > >>> .dynamic entry.
> > > > >>
> > > > >> The symbol version is needed to avoid problems due to copy relocations
> > > > >> if the symbol is referenced directly from the main program.  Without
> > > > >> that, the object could be truncated.  It's not a debugger
> > > > >> compatibility feature.
> > > > >
> > > > > Correct, but this violates *how* you're supposed to use _r_debug.
> > > >
> > > > If it is possible to link against it, we need to add the new symbol
> > > > version, in my opinion.
> > > >
> > > > > In the dynamic case it is different. The symbol should be looked up
> > > > > via DT_DEBUG only which always points to the library-local address
> > > > > of the data object (and the most recent version). In effect this
> > > > > bypasses the COPY relocation?
> > > >
> > > > How is this supposed to work if the dynamic linker does contain
> > > > DT_DEBUG?
> > > >
> > > > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > > > mapped at a random address even for ET_EXEC main programs, there must be
> > > > some other mechanism to locate it.
> > > >
> > > > Thanks,
> > > > Florian
> > > >
> > >
> > > I opened:
> > >
> > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > >
> > > with a proposal to extend struct r_debug for libraries loaded with
> > > dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> > > and GDB.
> >
> >
> > I have an updated set of changes which add a new dynamic entry
> > DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
> > implementation. I'm open to support different implementations.

Please send an email to gABI group:

https://groups.google.com/g/generic-abi

to add a new DT_XXX.   If it is impractical to add a new DT_XXX to gABI,
I propose DT_GNU_DEBUG:

https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2

to cover dlmopen and beyond.

> Let's work on it after glibc 2.34 is branched.
>
> > I found while working on this project that the clang linker adds 3 proprietary new
> > dynamic entries. So there is demand to make adding to the table easier.
> >
> > Daniel
>
>
>
> --
> H.J.



-- 
H.J.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Daniel Walker via Libc-alpha]

On Wed, Jul 28, 2021 at 08:15:10AM -0700, H.J. Lu wrote:
> On Tue, Jul 27, 2021 at 11:14 AM H.J. Lu <hjl.tools@gmail.com> wrote:
> >
> > On Tue, Jul 27, 2021 at 10:40 AM Daniel Walker <danielwa@cisco.com> wrote:
> > >
> > > On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> > > > On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> > > > <libc-alpha@sourceware.org> wrote:
> > > > >
> > > > > * Carlos O'Donell via Libc-alpha:
> > > > >
> > > > > >>> I'm not sure it would work to version _r_debug, since the debugger
> > > > > >>> is using DT_DEBUG and we only get to put one value in that
> > > > > >>> .dynamic entry.
> > > > > >>
> > > > > >> The symbol version is needed to avoid problems due to copy relocations
> > > > > >> if the symbol is referenced directly from the main program.  Without
> > > > > >> that, the object could be truncated.  It's not a debugger
> > > > > >> compatibility feature.
> > > > > >
> > > > > > Correct, but this violates *how* you're supposed to use _r_debug.
> > > > >
> > > > > If it is possible to link against it, we need to add the new symbol
> > > > > version, in my opinion.
> > > > >
> > > > > > In the dynamic case it is different. The symbol should be looked up
> > > > > > via DT_DEBUG only which always points to the library-local address
> > > > > > of the data object (and the most recent version). In effect this
> > > > > > bypasses the COPY relocation?
> > > > >
> > > > > How is this supposed to work if the dynamic linker does contain
> > > > > DT_DEBUG?
> > > > >
> > > > > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > > > > mapped at a random address even for ET_EXEC main programs, there must be
> > > > > some other mechanism to locate it.
> > > > >
> > > > > Thanks,
> > > > > Florian
> > > > >
> > > >
> > > > I opened:
> > > >
> > > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > > >
> > > > with a proposal to extend struct r_debug for libraries loaded with
> > > > dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> > > > and GDB.
> > >
> > >
> > > I have an updated set of changes which add a new dynamic entry
> > > DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
> > > implementation. I'm open to support different implementations.
> 
> Please send an email to gABI group:
> 
> https://groups.google.com/g/generic-abi
> 
> to add a new DT_XXX.   If it is impractical to add a new DT_XXX to gABI,
> I propose DT_GNU_DEBUG:
> 
> https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> 
> to cover dlmopen and beyond.
> 


The last time this was discussed I thought this was the gABI group. Someone had
said gABI was getting taken over by glibc.


Daniel

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[H.J. Lu via Libc-alpha]

On Wed, Jul 28, 2021 at 8:44 AM Daniel Walker <danielwa@cisco.com> wrote:
>
> On Wed, Jul 28, 2021 at 08:15:10AM -0700, H.J. Lu wrote:
> > On Tue, Jul 27, 2021 at 11:14 AM H.J. Lu <hjl.tools@gmail.com> wrote:
> > >
> > > On Tue, Jul 27, 2021 at 10:40 AM Daniel Walker <danielwa@cisco.com> wrote:
> > > >
> > > > On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> > > > > On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> > > > > <libc-alpha@sourceware.org> wrote:
> > > > > >
> > > > > > * Carlos O'Donell via Libc-alpha:
> > > > > >
> > > > > > >>> I'm not sure it would work to version _r_debug, since the debugger
> > > > > > >>> is using DT_DEBUG and we only get to put one value in that
> > > > > > >>> .dynamic entry.
> > > > > > >>
> > > > > > >> The symbol version is needed to avoid problems due to copy relocations
> > > > > > >> if the symbol is referenced directly from the main program.  Without
> > > > > > >> that, the object could be truncated.  It's not a debugger
> > > > > > >> compatibility feature.
> > > > > > >
> > > > > > > Correct, but this violates *how* you're supposed to use _r_debug.
> > > > > >
> > > > > > If it is possible to link against it, we need to add the new symbol
> > > > > > version, in my opinion.
> > > > > >
> > > > > > > In the dynamic case it is different. The symbol should be looked up
> > > > > > > via DT_DEBUG only which always points to the library-local address
> > > > > > > of the data object (and the most recent version). In effect this
> > > > > > > bypasses the COPY relocation?
> > > > > >
> > > > > > How is this supposed to work if the dynamic linker does contain
> > > > > > DT_DEBUG?
> > > > > >
> > > > > > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > > > > > mapped at a random address even for ET_EXEC main programs, there must be
> > > > > > some other mechanism to locate it.
> > > > > >
> > > > > > Thanks,
> > > > > > Florian
> > > > > >
> > > > >
> > > > > I opened:
> > > > >
> > > > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > > > >
> > > > > with a proposal to extend struct r_debug for libraries loaded with
> > > > > dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> > > > > and GDB.
> > > >
> > > >
> > > > I have an updated set of changes which add a new dynamic entry
> > > > DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
> > > > implementation. I'm open to support different implementations.
> >
> > Please send an email to gABI group:
> >
> > https://groups.google.com/g/generic-abi
> >
> > to add a new DT_XXX.   If it is impractical to add a new DT_XXX to gABI,
> > I propose DT_GNU_DEBUG:
> >
> > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> >
> > to cover dlmopen and beyond.
> >
>
>
> The last time this was discussed I thought this was the gABI group. Someone had
> said gABI was getting taken over by glibc.

There are 2 gABI groups:

1. OS independent gABI:

https://groups.google.com/g/generic-abi

2. Linux/GNU specific extensions to gABI:

https://sourceware.org/mailman/listinfo/gnu-gabi

We can try #1 first.

-- 
H.J.

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[Daniel Walker via Libc-alpha]

On Wed, Jul 28, 2021 at 10:14:41AM -0700, H.J. Lu wrote:
> On Wed, Jul 28, 2021 at 8:44 AM Daniel Walker <danielwa@cisco.com> wrote:
> >
> > On Wed, Jul 28, 2021 at 08:15:10AM -0700, H.J. Lu wrote:
> > > On Tue, Jul 27, 2021 at 11:14 AM H.J. Lu <hjl.tools@gmail.com> wrote:
> > > >
> > > > On Tue, Jul 27, 2021 at 10:40 AM Daniel Walker <danielwa@cisco.com> wrote:
> > > > >
> > > > > On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> > > > > > On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> > > > > > <libc-alpha@sourceware.org> wrote:
> > > > > > >
> > > > > > > * Carlos O'Donell via Libc-alpha:
> > > > > > >
> > > > > > > >>> I'm not sure it would work to version _r_debug, since the debugger
> > > > > > > >>> is using DT_DEBUG and we only get to put one value in that
> > > > > > > >>> .dynamic entry.
> > > > > > > >>
> > > > > > > >> The symbol version is needed to avoid problems due to copy relocations
> > > > > > > >> if the symbol is referenced directly from the main program.  Without
> > > > > > > >> that, the object could be truncated.  It's not a debugger
> > > > > > > >> compatibility feature.
> > > > > > > >
> > > > > > > > Correct, but this violates *how* you're supposed to use _r_debug.
> > > > > > >
> > > > > > > If it is possible to link against it, we need to add the new symbol
> > > > > > > version, in my opinion.
> > > > > > >
> > > > > > > > In the dynamic case it is different. The symbol should be looked up
> > > > > > > > via DT_DEBUG only which always points to the library-local address
> > > > > > > > of the data object (and the most recent version). In effect this
> > > > > > > > bypasses the COPY relocation?
> > > > > > >
> > > > > > > How is this supposed to work if the dynamic linker does contain
> > > > > > > DT_DEBUG?
> > > > > > >
> > > > > > > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > > > > > > mapped at a random address even for ET_EXEC main programs, there must be
> > > > > > > some other mechanism to locate it.
> > > > > > >
> > > > > > > Thanks,
> > > > > > > Florian
> > > > > > >
> > > > > >
> > > > > > I opened:
> > > > > >
> > > > > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > > > > >
> > > > > > with a proposal to extend struct r_debug for libraries loaded with
> > > > > > dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> > > > > > and GDB.
> > > > >
> > > > >
> > > > > I have an updated set of changes which add a new dynamic entry
> > > > > DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
> > > > > implementation. I'm open to support different implementations.
> > >
> > > Please send an email to gABI group:
> > >
> > > https://groups.google.com/g/generic-abi
> > >
> > > to add a new DT_XXX.   If it is impractical to add a new DT_XXX to gABI,
> > > I propose DT_GNU_DEBUG:
> > >
> > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > >
> > > to cover dlmopen and beyond.
> > >
> >
> >
> > The last time this was discussed I thought this was the gABI group. Someone had
> > said gABI was getting taken over by glibc.
> 
> There are 2 gABI groups:
> 
> 1. OS independent gABI:
> 
> https://groups.google.com/g/generic-abi
> 
> 2. Linux/GNU specific extensions to gABI:
> 
> https://sourceware.org/mailman/listinfo/gnu-gabi
> 
> We can try #1 first.

Do you want to drive this, or should I ? It looks like you know the people
involved better than I do.

Daniel

Re: [RFC PATCH 3/3] add r_debug multiple namespaces support

[H.J. Lu via Libc-alpha]

On Wed, Jul 28, 2021 at 12:02 PM Daniel Walker <danielwa@cisco.com> wrote:
>
> On Wed, Jul 28, 2021 at 10:14:41AM -0700, H.J. Lu wrote:
> > On Wed, Jul 28, 2021 at 8:44 AM Daniel Walker <danielwa@cisco.com> wrote:
> > >
> > > On Wed, Jul 28, 2021 at 08:15:10AM -0700, H.J. Lu wrote:
> > > > On Tue, Jul 27, 2021 at 11:14 AM H.J. Lu <hjl.tools@gmail.com> wrote:
> > > > >
> > > > > On Tue, Jul 27, 2021 at 10:40 AM Daniel Walker <danielwa@cisco.com> wrote:
> > > > > >
> > > > > > On Fri, Jul 23, 2021 at 04:38:33PM -0700, H.J. Lu wrote:
> > > > > > > On Mon, Jun 29, 2020 at 2:49 AM Florian Weimer via Libc-alpha
> > > > > > > <libc-alpha@sourceware.org> wrote:
> > > > > > > >
> > > > > > > > * Carlos O'Donell via Libc-alpha:
> > > > > > > >
> > > > > > > > >>> I'm not sure it would work to version _r_debug, since the debugger
> > > > > > > > >>> is using DT_DEBUG and we only get to put one value in that
> > > > > > > > >>> .dynamic entry.
> > > > > > > > >>
> > > > > > > > >> The symbol version is needed to avoid problems due to copy relocations
> > > > > > > > >> if the symbol is referenced directly from the main program.  Without
> > > > > > > > >> that, the object could be truncated.  It's not a debugger
> > > > > > > > >> compatibility feature.
> > > > > > > > >
> > > > > > > > > Correct, but this violates *how* you're supposed to use _r_debug.
> > > > > > > >
> > > > > > > > If it is possible to link against it, we need to add the new symbol
> > > > > > > > version, in my opinion.
> > > > > > > >
> > > > > > > > > In the dynamic case it is different. The symbol should be looked up
> > > > > > > > > via DT_DEBUG only which always points to the library-local address
> > > > > > > > > of the data object (and the most recent version). In effect this
> > > > > > > > > bypasses the COPY relocation?
> > > > > > > >
> > > > > > > > How is this supposed to work if the dynamic linker does contain
> > > > > > > > DT_DEBUG?
> > > > > > > >
> > > > > > > > I only observe DT_DEBUG in PIE binaries, but since the dynamic loader is
> > > > > > > > mapped at a random address even for ET_EXEC main programs, there must be
> > > > > > > > some other mechanism to locate it.
> > > > > > > >
> > > > > > > > Thanks,
> > > > > > > > Florian
> > > > > > > >
> > > > > > >
> > > > > > > I opened:
> > > > > > >
> > > > > > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > > > > > >
> > > > > > > with a proposal to extend struct r_debug for libraries loaded with
> > > > > > > dlmopen,  I'd like to resolve it for the next releases of glibc, binutils
> > > > > > > and GDB.
> > > > > >
> > > > > >
> > > > > > I have an updated set of changes which add a new dynamic entry
> > > > > > DT_DEBUG_DLMOPEN, which was recommended by Carlos. We're still testing the
> > > > > > implementation. I'm open to support different implementations.
> > > >
> > > > Please send an email to gABI group:
> > > >
> > > > https://groups.google.com/g/generic-abi
> > > >
> > > > to add a new DT_XXX.   If it is impractical to add a new DT_XXX to gABI,
> > > > I propose DT_GNU_DEBUG:
> > > >
> > > > https://gitlab.com/x86-psABIs/Linux-ABI/-/issues/2
> > > >
> > > > to cover dlmopen and beyond.
> > > >
> > >
> > >
> > > The last time this was discussed I thought this was the gABI group. Someone had
> > > said gABI was getting taken over by glibc.
> >
> > There are 2 gABI groups:
> >
> > 1. OS independent gABI:
> >
> > https://groups.google.com/g/generic-abi
> >
> > 2. Linux/GNU specific extensions to gABI:
> >
> > https://sourceware.org/mailman/listinfo/gnu-gabi
> >
> > We can try #1 first.
>
> Do you want to drive this, or should I ? It looks like you know the people
> involved better than I do.
>

https://groups.google.com/g/generic-abi/c/1ngxmSwrafc

-- 
H.J.

RFC: Add DT_GNU_DEBUG

[H.J. Lu via Libc-alpha]

On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> > Do you want to drive this, or should I ? It looks like you know the people
> > involved better than I do.
> >
>
> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
>

I don't think the gABI community is interested in a new debug dynamic
tag.  I propose DT_GNU_DEBUG:

#define DT_GNU_DEBUG   0x6ffffff8

for the address of a pointer which will be filled by the dynamic
linker.  Linker should
add the DT_GNU_DEBUG entry to executable's dynamic section.

BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
relocation, like DT_DEBUG.

-- 
H.J.

Re: RFC: Add DT_GNU_DEBUG

[Florian Weimer via Libc-alpha]

* H. J. Lu:

> On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>> > Do you want to drive this, or should I ? It looks like you know the people
>> > involved better than I do.
>> >
>>
>> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
>>
>
> I don't think the gABI community is interested in a new debug dynamic
> tag.  I propose DT_GNU_DEBUG:
>
> #define DT_GNU_DEBUG   0x6ffffff8
>
> for the address of a pointer which will be filled by the dynamic
> linker.  Linker should
> add the DT_GNU_DEBUG entry to executable's dynamic section.
>
> BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
> relocation, like DT_DEBUG.

What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
to by DT_DEBUG?

Perhaps the gABI folks are interested in that, too?  I think it's worth
a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
GNU ABI.

Thanks,
Florian

Re: RFC: Add DT_GNU_DEBUG

[H.J. Lu via Libc-alpha]

On Sun, Aug 1, 2021 at 10:22 PM Florian Weimer <fweimer@redhat.com> wrote:
>
> * H. J. Lu:
>
> > On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >> > Do you want to drive this, or should I ? It looks like you know the people
> >> > involved better than I do.
> >> >
> >>
> >> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
> >>
> >
> > I don't think the gABI community is interested in a new debug dynamic
> > tag.  I propose DT_GNU_DEBUG:
> >
> > #define DT_GNU_DEBUG   0x6ffffff8
> >
> > for the address of a pointer which will be filled by the dynamic
> > linker.  Linker should
> > add the DT_GNU_DEBUG entry to executable's dynamic section.
> >
> > BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
> > relocation, like DT_DEBUG.
>
> What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
> to by DT_DEBUG?

It works if we don't need to support static executables.

> Perhaps the gABI folks are interested in that, too?  I think it's worth
> a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
> GNU ABI.

I did.  I didn't get any feedback.

-- 
H.J.

Re: RFC: Add DT_GNU_DEBUG

[Daniel Walker via Libc-alpha]

On Mon, Aug 02, 2021 at 06:10:55AM -0700, H.J. Lu wrote:
> On Sun, Aug 1, 2021 at 10:22 PM Florian Weimer <fweimer@redhat.com> wrote:
> >
> > * H. J. Lu:
> >
> > > On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> > >> > Do you want to drive this, or should I ? It looks like you know the people
> > >> > involved better than I do.
> > >> >
> > >>
> > >> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
> > >>
> > >
> > > I don't think the gABI community is interested in a new debug dynamic
> > > tag.  I propose DT_GNU_DEBUG:
> > >
> > > #define DT_GNU_DEBUG   0x6ffffff8
> > >
> > > for the address of a pointer which will be filled by the dynamic
> > > linker.  Linker should
> > > add the DT_GNU_DEBUG entry to executable's dynamic section.
> > >
> > > BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
> > > relocation, like DT_DEBUG.
> >
> > What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
> > to by DT_DEBUG?
> 
> It works if we don't need to support static executables.
> 
> > Perhaps the gABI folks are interested in that, too?  I think it's worth
> > a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
> > GNU ABI.
> 
> I did.  I didn't get any feedback.

So no feedback equal "not interested" ?

Daniel

Re: RFC: Add DT_GNU_DEBUG

[H.J. Lu via Libc-alpha]

On Tue, Aug 3, 2021 at 9:39 AM Daniel Walker <danielwa@cisco.com> wrote:
>
> On Mon, Aug 02, 2021 at 06:10:55AM -0700, H.J. Lu wrote:
> > On Sun, Aug 1, 2021 at 10:22 PM Florian Weimer <fweimer@redhat.com> wrote:
> > >
> > > * H. J. Lu:
> > >
> > > > On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> > > >> > Do you want to drive this, or should I ? It looks like you know the people
> > > >> > involved better than I do.
> > > >> >
> > > >>
> > > >> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
> > > >>
> > > >
> > > > I don't think the gABI community is interested in a new debug dynamic
> > > > tag.  I propose DT_GNU_DEBUG:
> > > >
> > > > #define DT_GNU_DEBUG   0x6ffffff8
> > > >
> > > > for the address of a pointer which will be filled by the dynamic
> > > > linker.  Linker should
> > > > add the DT_GNU_DEBUG entry to executable's dynamic section.
> > > >
> > > > BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
> > > > relocation, like DT_DEBUG.
> > >
> > > What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
> > > to by DT_DEBUG?
> >
> > It works if we don't need to support static executables.

Given that we export _r_debug and some programs, like GNAT, use it,
we should keep and fix _r_debug.  We should also make the new interface
available for these programs and include the structure size in the new
interface.

> >
> > > Perhaps the gABI folks are interested in that, too?  I think it's worth
> > > a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
> > > GNU ABI.
> >
> > I did.  I didn't get any feedback.
>
> So no feedback equal "not interested" ?

I'd like to resolve this issue for glibc 2.35.  We need to move forward with
a new DT_XXX.  We can't wait too long.


-- 
H.J.

Re: RFC: Add DT_GNU_DEBUG

[Florian Weimer via Libc-alpha]

* Daniel Walker:

> On Mon, Aug 02, 2021 at 06:10:55AM -0700, H.J. Lu wrote:
>> On Sun, Aug 1, 2021 at 10:22 PM Florian Weimer <fweimer@redhat.com> wrote:
>> >
>> > * H. J. Lu:
>> >
>> > > On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>> > >> > Do you want to drive this, or should I ? It looks like you know the people
>> > >> > involved better than I do.
>> > >> >
>> > >>
>> > >> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
>> > >>
>> > >
>> > > I don't think the gABI community is interested in a new debug dynamic
>> > > tag.  I propose DT_GNU_DEBUG:
>> > >
>> > > #define DT_GNU_DEBUG   0x6ffffff8
>> > >
>> > > for the address of a pointer which will be filled by the dynamic
>> > > linker.  Linker should
>> > > add the DT_GNU_DEBUG entry to executable's dynamic section.
>> > >
>> > > BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
>> > > relocation, like DT_DEBUG.
>> >
>> > What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
>> > to by DT_DEBUG?
>> 
>> It works if we don't need to support static executables.
>> 
>> > Perhaps the gABI folks are interested in that, too?  I think it's worth
>> > a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
>> > GNU ABI.
>> 
>> I did.  I didn't get any feedback.
>
> So no feedback equal "not interested" ?

I think the issue is that the struct already has a version field.

It's only a problem for us because we have architectures with copy
relocations *and* have exposed _r_debug as a public symbol.  However, we
can do what we did for _sys_errlist in the past (new symbol versions for
each size change), so it's not a real blocker.

I suppose the easiest way forward would be to grow _r_debug this way,
bump the version field past Solaris' version, add the Solaris members
(possibly with dummy values), and add our own new stuff afterwards.

Thanks,
Florian

Re: RFC: Add DT_GNU_DEBUG

[H.J. Lu via Libc-alpha]

On Tue, Aug 3, 2021 at 11:12 AM Florian Weimer <fweimer@redhat.com> wrote:
>
> * Daniel Walker:
>
> > On Mon, Aug 02, 2021 at 06:10:55AM -0700, H.J. Lu wrote:
> >> On Sun, Aug 1, 2021 at 10:22 PM Florian Weimer <fweimer@redhat.com> wrote:
> >> >
> >> > * H. J. Lu:
> >> >
> >> > > On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >> > >> > Do you want to drive this, or should I ? It looks like you know the people
> >> > >> > involved better than I do.
> >> > >> >
> >> > >>
> >> > >> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
> >> > >>
> >> > >
> >> > > I don't think the gABI community is interested in a new debug dynamic
> >> > > tag.  I propose DT_GNU_DEBUG:
> >> > >
> >> > > #define DT_GNU_DEBUG   0x6ffffff8
> >> > >
> >> > > for the address of a pointer which will be filled by the dynamic
> >> > > linker.  Linker should
> >> > > add the DT_GNU_DEBUG entry to executable's dynamic section.
> >> > >
> >> > > BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
> >> > > relocation, like DT_DEBUG.
> >> >
> >> > What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
> >> > to by DT_DEBUG?
> >>
> >> It works if we don't need to support static executables.
> >>
> >> > Perhaps the gABI folks are interested in that, too?  I think it's worth
> >> > a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
> >> > GNU ABI.
> >>
> >> I did.  I didn't get any feedback.
> >
> > So no feedback equal "not interested" ?
>
> I think the issue is that the struct already has a version field.
>
> It's only a problem for us because we have architectures with copy
> relocations *and* have exposed _r_debug as a public symbol.  However, we
> can do what we did for _sys_errlist in the past (new symbol versions for
> each size change), so it's not a real blocker.

No, copy relocation doesn't work:

https://sourceware.org/bugzilla/show_bug.cgi?id=28130

I want to get rid of copy relocation and keep _r_debug only for existing
broken binaries.   The new interface will provide an access function to
get the address of internal data in ld.so.

> I suppose the easiest way forward would be to grow _r_debug this way,
> bump the version field past Solaris' version, add the Solaris members
> (possibly with dummy values), and add our own new stuff afterwards.
>
> Thanks,
> Florian
>


-- 
H.J.

Re: RFC: Add DT_GNU_DEBUG

[Daniel Walker via Libc-alpha]

On Tue, Aug 03, 2021 at 11:08:35AM -0700, H.J. Lu wrote:
> On Tue, Aug 3, 2021 at 9:39 AM Daniel Walker <danielwa@cisco.com> wrote:
> >
> > On Mon, Aug 02, 2021 at 06:10:55AM -0700, H.J. Lu wrote:
> > > On Sun, Aug 1, 2021 at 10:22 PM Florian Weimer <fweimer@redhat.com> wrote:
> > > >
> > > > * H. J. Lu:
> > > >
> > > > > On Wed, Jul 28, 2021 at 1:04 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> > > > >> > Do you want to drive this, or should I ? It looks like you know the people
> > > > >> > involved better than I do.
> > > > >> >
> > > > >>
> > > > >> https://groups.google.com/g/generic-abi/c/1ngxmSwrafc
> > > > >>
> > > > >
> > > > > I don't think the gABI community is interested in a new debug dynamic
> > > > > tag.  I propose DT_GNU_DEBUG:
> > > > >
> > > > > #define DT_GNU_DEBUG   0x6ffffff8
> > > > >
> > > > > for the address of a pointer which will be filled by the dynamic
> > > > > linker.  Linker should
> > > > > add the DT_GNU_DEBUG entry to executable's dynamic section.
> > > > >
> > > > > BTW, we have a choice.  DT_GNU_DEBUG can be readonly or readonly after
> > > > > relocation, like DT_DEBUG.
> > > >
> > > > What about adding DT_DEBUG_SIZE, specifying the size of the data pointed
> > > > to by DT_DEBUG?
> > >
> > > It works if we don't need to support static executables.
> 
> Given that we export _r_debug and some programs, like GNAT, use it,
> we should keep and fix _r_debug.  We should also make the new interface
> available for these programs and include the structure size in the new
> interface.
> 
> > >
> > > > Perhaps the gABI folks are interested in that, too?  I think it's worth
> > > > a try.  If the answer is “no”, we can still add DT_GNU_DEBUG_SIZE to the
> > > > GNU ABI.
> > >
> > > I did.  I didn't get any feedback.
> >
> > So no feedback equal "not interested" ?
> 
> I'd like to resolve this issue for glibc 2.35.  We need to move forward with
> a new DT_XXX.  We can't wait too long.

What sort of response are you specifically looking for ?

Daniel

Re: RFC: Add DT_GNU_DEBUG

[Florian Weimer via Libc-alpha]

* H. J. Lu:

> No, copy relocation doesn't work:
>
> https://sourceware.org/bugzilla/show_bug.cgi?id=28130
>
> I want to get rid of copy relocation and keep _r_debug only for existing
> broken binaries.   The new interface will provide an access function to
> get the address of internal data in ld.so.

Surely we can update both copies?!  They exist today, so I don't expect
new problems from a tools perspective.

Thanks,
Florian

Re: RFC: Add DT_GNU_DEBUG

[H.J. Lu via Libc-alpha]

On Tue, Aug 3, 2021 at 1:13 PM Florian Weimer <fweimer@redhat.com> wrote:
>
> * H. J. Lu:
>
> > No, copy relocation doesn't work:
> >
> > https://sourceware.org/bugzilla/show_bug.cgi?id=28130
> >
> > I want to get rid of copy relocation and keep _r_debug only for existing
> > broken binaries.   The new interface will provide an access function to
> > get the address of internal data in ld.so.
>
> Surely we can update both copies?!  They exist today, so I don't expect
> new problems from a tools perspective.
>

Why should we do that?  We don't do that for errno.

-- 
H.J.

RFC: 2 choices of DT_XXX for dlmopen

[H.J. Lu via Libc-alpha]

We need a new DT_XXX to support dlmopen.  We have 2 choices:

1. Similar to DT_DEBUG.  Linker will allocate a new DT_XXX and
ld.so will fill it with the address of the new debug data structure for
dlmopen.
2. Similar to DT_MIPS_RLD_MAP_REL/DT_MIPS_RLD_MAP.
Linker will allocate a space for a pointer, a new DT_XXX and fill
the DT_XXX entry with the address of the pointer.   ld.so will update
the pointer with the address of the new debug data structure for
dlmopen.

#1 is the most straightforward to implement.  #2 is compatible with
the current MIPS implementation.

Does anyone have any preferences?

-- 
H.J.

Re: RFC: 2 choices of DT_XXX for dlmopen

[Daniel Walker via Libc-alpha]

On Mon, Aug 09, 2021 at 07:32:26AM -0700, H.J. Lu wrote:
> We need a new DT_XXX to support dlmopen.  We have 2 choices:
> 
> 1. Similar to DT_DEBUG.  Linker will allocate a new DT_XXX and
> ld.so will fill it with the address of the new debug data structure for
> dlmopen.
> 2. Similar to DT_MIPS_RLD_MAP_REL/DT_MIPS_RLD_MAP.
> Linker will allocate a space for a pointer, a new DT_XXX and fill
> the DT_XXX entry with the address of the pointer.   ld.so will update
> the pointer with the address of the new debug data structure for
> dlmopen.
> 
> #1 is the most straightforward to implement.  #2 is compatible with
> the current MIPS implementation.
> 
> Does anyone have any preferences?


I have #1 fully implemented already.

Daniel

[PATCH] Extend struct r_debug to support multiple namespaces

[H.J. Lu via Libc-alpha]

On Mon, Aug 09, 2021 at 10:16:53AM -0700, Daniel Walker wrote:
> On Mon, Aug 09, 2021 at 07:32:26AM -0700, H.J. Lu wrote:
> > We need a new DT_XXX to support dlmopen.  We have 2 choices:
> > 
> > 1. Similar to DT_DEBUG.  Linker will allocate a new DT_XXX and
> > ld.so will fill it with the address of the new debug data structure for
> > dlmopen.
> > 2. Similar to DT_MIPS_RLD_MAP_REL/DT_MIPS_RLD_MAP.
> > Linker will allocate a space for a pointer, a new DT_XXX and fill
> > the DT_XXX entry with the address of the pointer.   ld.so will update
> > the pointer with the address of the new debug data structure for
> > dlmopen.
> > 
> > #1 is the most straightforward to implement.  #2 is compatible with
> > the current MIPS implementation.
> > 
> > Does anyone have any preferences?
> 
> 
> I have #1 fully implemented already.
> 

Here is a slightly different approach with DT_DEBUGSZ.  The corresponding
linker patch is at

https://sourceware.org/pipermail/binutils/2021-August/117706.html

H.J.
---
Glibc does not provide an interface for debugger to access libraries
loaded in multiple namespaces via dlmopen.

The current rtld-debugger interface is described in the file:

elf/rtld-debugger-interface.txt

under the "Standard debugger interface" heading.  This interface only
provides access to the first link-map (LM_ID_BASE).

Based on the patch from Conan C Huang <conhuang@cisco.com>:

https://sourceware.org/pipermail/libc-alpha/2020-June/115448.html

1. Extend struct r_debug into a linked-list, where each element correlates
to an unique namespace.
2. Add a new dynamic tag, DT_DEBUGSZ,

 #define DT_DEBUGSZ 0x6ffffff8

which records the size of the extended struct r_debug.
3. For executables with a DT_DEBUGSZ dynamic tag, debugger (GDB) can
traverse r_debug linked-list in the extended struct r_debug from the
DT_DEBUG dynamic tag to retrieve information for all loaded namespaces.
4. Provide the compatibility symbol, _r_debug, for programs which
reference _r_debug and update the copy of _r_debug in executable if
copy relocation is used.
---
 config.h.in                     |  4 +++
 configure                       | 43 +++++++++++++++++++++++
 configure.ac                    | 12 +++++++
 csu/Makefile                    |  3 ++
 csu/libc-start.c                |  2 +-
 csu/rtld-sizes.sym              |  4 +++
 elf/Makefile                    | 24 +++++++++++--
 elf/dl-close.c                  |  7 +++-
 elf/dl-debug-symbols-gen.c      | 24 +++++++++++++
 elf/dl-debug-symbols.S          | 34 +++++++++++++++++++
 elf/dl-debug.c                  | 35 +++++++++++++------
 elf/dl-load.c                   |  7 +++-
 elf/dl-open.c                   | 19 +++++++----
 elf/dl-reloc-static-pie.c       |  8 ++++-
 elf/elf.h                       |  1 +
 elf/link.h                      | 13 ++++++-
 elf/rtld-debugger-interface.txt | 12 +++++++
 elf/rtld.c                      | 10 ++++--
 elf/tst-_r_gnu_debug.c          | 60 +++++++++++++++++++++++++++++++++
 include/link.h                  |  4 +++
 sysdeps/generic/ldsodefs.h      |  6 ++--
 21 files changed, 304 insertions(+), 28 deletions(-)
 create mode 100644 csu/rtld-sizes.sym
 create mode 100644 elf/dl-debug-symbols-gen.c
 create mode 100644 elf/dl-debug-symbols.S
 create mode 100644 elf/tst-_r_gnu_debug.c

diff --git a/config.h.in b/config.h.in
index 0d92504f65..f05cf57f38 100644
--- a/config.h.in
+++ b/config.h.in
@@ -59,6 +59,10 @@
 /* Define if the linker supports the -z combreloc option.  */
 #undef	HAVE_Z_COMBRELOC
 
+/* Define to 1 if the linker supports the -z dt-debugsz option to emit
+   the DT_DEBUGSZ dynamic tag in executables.  */
+#undef HAVE_Z_DT_DEBUGSZ
+
 /* Define if _rtld_local structure should be forced into .sdata section.  */
 #undef	HAVE_SDATA_SECTION
 
diff --git a/configure b/configure
index 7272fbf6ea..6d06a1c87b 100755
--- a/configure
+++ b/configure
@@ -6003,6 +6003,49 @@ $as_echo "$libc_linker_feature" >&6; }
 config_vars="$config_vars
 have-no-dynamic-linker = $libc_cv_no_dynamic_linker"
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for linker that supports -z dt-debugsz" >&5
+$as_echo_n "checking for linker that supports -z dt-debugsz... " >&6; }
+libc_linker_feature=no
+if test x"$gnu_ld" = x"yes"; then
+  libc_linker_check=`$LD -v --help 2>/dev/null | grep "\-z dt-debugsz"`
+  if test -n "$libc_linker_check"; then
+    cat > conftest.c <<EOF
+int _start (void) { return 42; }
+EOF
+    if { ac_try='${CC-cc} $CFLAGS $CPPFLAGS $LDFLAGS $no_ssp
+				-Wl,-z,dt-debugsz -nostdlib -nostartfiles
+				-fPIC -shared -o conftest.so conftest.c
+				1>&5'
+  { { eval echo "\"\$as_me\":${as_lineno-$LINENO}: \"$ac_try\""; } >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }
+    then
+      libc_linker_feature=yes
+    fi
+    rm -f conftest*
+  fi
+fi
+if test $libc_linker_feature = yes; then
+  libc_cv_z_dt_debugsz=yes
+else
+  libc_cv_z_dt_debugsz=no
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $libc_linker_feature" >&5
+$as_echo "$libc_linker_feature" >&6; }
+config_vars="$config_vars
+have-z-dt-debugsz = $libc_cv_z_dt_debugsz"
+if test $libc_cv_z_dt_debugsz = yes; then
+  have_z_dt_debugsz=1
+else
+  have_z_dt_debugsz=0
+fi
+cat >>confdefs.h <<_ACEOF
+#define HAVE_Z_DT_DEBUGSZ $have_z_dt_debugsz
+_ACEOF
+
+
 { $as_echo "$as_me:${as_lineno-$LINENO}: checking for -static-pie" >&5
 $as_echo_n "checking for -static-pie... " >&6; }
 if ${libc_cv_static_pie+:} false; then :
diff --git a/configure.ac b/configure.ac
index af47cd51e6..c5ffa381eb 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1343,6 +1343,18 @@ LIBC_LINKER_FEATURE([--no-dynamic-linker],
 		    [libc_cv_no_dynamic_linker=no])
 LIBC_CONFIG_VAR([have-no-dynamic-linker], [$libc_cv_no_dynamic_linker])
 
+LIBC_LINKER_FEATURE([-z dt-debugsz],
+		    [-Wl,-z,dt-debugsz],
+		    [libc_cv_z_dt_debugsz=yes],
+		    [libc_cv_z_dt_debugsz=no])
+LIBC_CONFIG_VAR([have-z-dt-debugsz], [$libc_cv_z_dt_debugsz])
+if test $libc_cv_z_dt_debugsz = yes; then
+  have_z_dt_debugsz=1
+else
+  have_z_dt_debugsz=0
+fi
+AC_DEFINE_UNQUOTED([HAVE_Z_DT_DEBUGSZ], [$have_z_dt_debugsz])
+
 AC_CACHE_CHECK(for -static-pie, libc_cv_static_pie, [dnl
 LIBC_TRY_CC_OPTION([-static-pie],
 		   [libc_cv_static_pie=yes],
diff --git a/csu/Makefile b/csu/Makefile
index 3054329cea..e2390e4a7d 100644
--- a/csu/Makefile
+++ b/csu/Makefile
@@ -88,6 +88,9 @@ endif
 before-compile += $(objpfx)abi-tag.h
 generated += abi-tag.h
 
+# Put it here to generate it earlier.
+gen-as-const-headers += rtld-sizes.sym
+
 # These are the special initializer/finalizer files.  They are always the
 # first and last file in the link.  crti.o ... crtn.o define the global
 # "functions" _init and _fini to run the .init and .fini sections.
diff --git a/csu/libc-start.c b/csu/libc-start.c
index 0350b006fd..8cfa68a057 100644
--- a/csu/libc-start.c
+++ b/csu/libc-start.c
@@ -403,7 +403,7 @@ LIBC_START_MAIN (int (*main) (int, char **, char ** MAIN_AUXVEC_DECL),
 #endif
 
 #ifndef SHARED
-  _dl_debug_initialize (0, LM_ID_BASE);
+  _dl_debug_initialize (0, LM_ID_BASE, NULL);
 #endif
 
   __libc_start_call_main (main, argc, argv MAIN_AUXVEC_PARAM);
diff --git a/csu/rtld-sizes.sym b/csu/rtld-sizes.sym
new file mode 100644
index 0000000000..2583e58b19
--- /dev/null
+++ b/csu/rtld-sizes.sym
@@ -0,0 +1,4 @@
+#include <link.h>
+
+--
+COMPAT_R_DEBUG_SIZE	offsetof (struct r_debug, r_ldbase) + sizeof (ElfW(Addr))
diff --git a/elf/Makefile b/elf/Makefile
index d05f410592..c97c9f8468 100644
--- a/elf/Makefile
+++ b/elf/Makefile
@@ -35,7 +35,8 @@ dl-routines	= $(addprefix dl-,load lookup object reloc deps \
 				  execstack open close trampoline \
 				  exception sort-maps lookup-direct \
 				  call-libc-early-init write \
-				  thread_gscope_wait tls_init_tp)
+				  thread_gscope_wait tls_init_tp \
+				  debug-symbols)
 ifeq (yes,$(use-ldconfig))
 dl-routines += dl-cache
 endif
@@ -224,7 +225,7 @@ tests += restest1 preloadtest loadfail multiload origtest resolvfail \
 	 tst-tls-ie tst-tls-ie-dlmopen argv0test \
 	 tst-glibc-hwcaps tst-glibc-hwcaps-prepend tst-glibc-hwcaps-mask \
 	 tst-tls20 tst-tls21 tst-dlmopen-dlerror tst-dlmopen-gethostbyname \
-	 tst-dl-is_dso
+	 tst-dl-is_dso tst-_r_gnu_debug
 #	 reldep9
 tests-internal += loadtest unload unload2 circleload1 \
 	 neededtest neededtest2 neededtest3 neededtest4 \
@@ -672,6 +673,21 @@ LC_ALL=C sed $(ldd-rewrite) < $< \
 endef
 endif
 
+ifeq ($(build-shared),yes)
+generated += dl-debug-compat-symbols.os dl-debug-compat-symbols.o
+
+libof-dl-debug-compat-symbols = rtld
+
+$(objpfx)dl-debug-compat-symbols.os: dl-debug-symbols-gen.c
+	$(compile-command.c) -S
+
+$(objpfx)dl-debug-compat-symbols.o: dl-debug-symbols-gen.c
+	$(compile-command.c) -S
+
+$(objpfx)dl-debug-symbols.os: $(objpfx)dl-debug-compat-symbols.os
+$(objpfx)dl-debug-symbols.o: $(objpfx)dl-debug-compat-symbols.o
+endif
+
 $(objpfx)ldd: ldd.bash.in $(common-objpfx)soversions.mk \
 	      $(common-objpfx)config.make
 	$(gen-ldd)
@@ -1906,3 +1922,7 @@ $(objpfx)tst-getauxval-static.out: $(objpfx)tst-auxvalmod.so
 tst-getauxval-static-ENV = LD_LIBRARY_PATH=$(objpfx):$(common-objpfx)
 
 $(objpfx)tst-dlmopen-gethostbyname.out: $(objpfx)tst-dlmopen-gethostbyname-mod.so
+
+ifeq ($(have-z-dt-debugsz),yes)
+LDFLAGS-tst-_r_gnu_debug += -Wl,-z,dt-debugsz
+endif
diff --git a/elf/dl-close.c b/elf/dl-close.c
index f39001cab9..647a5fafe1 100644
--- a/elf/dl-close.c
+++ b/elf/dl-close.c
@@ -500,8 +500,11 @@ _dl_close_worker (struct link_map *map, bool force)
 #endif
 
   /* Notify the debugger we are about to remove some loaded objects.  */
-  struct r_debug *r = _dl_debug_initialize (0, nsid);
+  struct r_debug *r_debug_compat;
+  struct r_debug *r = _dl_debug_initialize (0, nsid, &r_debug_compat);
   r->r_state = RT_DELETE;
+  if (r_debug_compat != NULL)
+    r_debug_compat->r_state = RT_DELETE;
   _dl_debug_state ();
   LIBC_PROBE (unmap_start, 2, nsid, r);
 
@@ -821,6 +824,8 @@ _dl_close_worker (struct link_map *map, bool force)
 
   /* Notify the debugger those objects are finalized and gone.  */
   r->r_state = RT_CONSISTENT;
+  if (r_debug_compat != NULL)
+    r_debug_compat->r_state = RT_CONSISTENT;
   _dl_debug_state ();
   LIBC_PROBE (unmap_complete, 2, nsid, r);
 
diff --git a/elf/dl-debug-symbols-gen.c b/elf/dl-debug-symbols-gen.c
new file mode 100644
index 0000000000..a8d2aea75d
--- /dev/null
+++ b/elf/dl-debug-symbols-gen.c
@@ -0,0 +1,24 @@
+/* Generate the _r_gnu_debug symbol used to communicate dynamic linker
+   state to the debugger at runtime.
+   Copyright (C) 2021 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <link.h>
+
+/* This structure communicates dl state to the debugger.  The debugger
+   finds it via the DT_DEBUG entry in the dynamic section.  */
+struct r_debug _r_gnu_debug;
diff --git a/elf/dl-debug-symbols.S b/elf/dl-debug-symbols.S
new file mode 100644
index 0000000000..b43a2c310d
--- /dev/null
+++ b/elf/dl-debug-symbols.S
@@ -0,0 +1,34 @@
+/* Define symbols used to communicate dynamic linker state to the
+   debugger at runtime.
+   Copyright (C) 2021 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <rtld-sizes.h>
+
+#ifdef SHARED
+# include "dl-debug-compat-symbols.os"
+#else
+# include "dl-debug-compat-symbols.o"
+#endif
+
+/* Define the compatibility symbol, _r_debug, used to communicate
+   dynamic linker state to the debugger.  The debugger normally
+   finds it via the DT_DEBUG entry in the dynamic section, but in a
+   statically-linked program there is no dynamic section for the
+   debugger to examine and it looks for this particular symbol name.
+   This must be done after _r_gnu_debug has been defined first.  */
+declare_object_symbol_alias (_r_debug, _r_gnu_debug, COMPAT_R_DEBUG_SIZE);
diff --git a/elf/dl-debug.c b/elf/dl-debug.c
index 2cd5f09753..e7d634e237 100644
--- a/elf/dl-debug.c
+++ b/elf/dl-debug.c
@@ -30,34 +30,47 @@ extern const int verify_link_map_members[(VERIFY_MEMBER (l_addr)
 					  && VERIFY_MEMBER (l_prev))
 					 ? 1 : -1];
 
-/* This structure communicates dl state to the debugger.  The debugger
-   normally finds it via the DT_DEBUG entry in the dynamic section, but in
-   a statically-linked program there is no dynamic section for the debugger
-   to examine and it looks for this particular symbol name.  */
-struct r_debug _r_debug;
-
+/* The compatibility symbol, _r_debug, which can have a different
+   address from _r_gnu_debug at run-time due to copy relocation.  */
+extern struct r_debug _r_debug;
 
 /* Initialize _r_debug if it has not already been done.  The argument is
    the run-time load address of the dynamic linker, to be put in
    _r_debug.r_ldbase.  Returns the address of _r_debug.  */
 
 struct r_debug *
-_dl_debug_initialize (ElfW(Addr) ldbase, Lmid_t ns)
+_dl_debug_initialize (ElfW(Addr) ldbase, Lmid_t ns,
+		      struct r_debug ** _r_debug_compat_p)
 {
-  struct r_debug *r;
+  struct r_debug *r_debug_compat = NULL;
+  struct r_debug *r, *rp;
 
   if (ns == LM_ID_BASE)
-    r = &_r_debug;
+    {
+      r = &_r_gnu_debug;
+      if (r != &_r_debug)
+	r_debug_compat = &_r_debug;
+    }
   else
-    r = &GL(dl_ns)[ns]._ns_debug;
+    {
+      r = &GL(dl_ns)[ns]._ns_debug;
+      rp = &GL(dl_ns)[ns - 1]._ns_debug;
+      rp->next = r;
+      if (ns - 1 == LM_ID_BASE)
+        _r_gnu_debug.next = r;
+    }
+
+  if (_r_debug_compat_p != NULL)
+    *_r_debug_compat_p = r_debug_compat;
 
   if (r->r_map == NULL || ldbase != 0)
     {
       /* Tell the debugger where to find the map of loaded objects.  */
       r->r_version = 1	/* R_DEBUG_VERSION XXX */;
-      r->r_ldbase = ldbase ?: _r_debug.r_ldbase;
+      r->r_ldbase = ldbase ?: _r_gnu_debug.r_ldbase;
       r->r_map = (void *) GL(dl_ns)[ns]._ns_loaded;
       r->r_brk = (ElfW(Addr)) &_dl_debug_state;
+      r->next = NULL;
     }
 
   return r;
diff --git a/elf/dl-load.c b/elf/dl-load.c
index 650e4edc35..370f56a6d6 100644
--- a/elf/dl-load.c
+++ b/elf/dl-load.c
@@ -949,7 +949,8 @@ _dl_map_object_from_fd (const char *name, const char *origname, int fd,
   /* Initialize to keep the compiler happy.  */
   const char *errstring = NULL;
   int errval = 0;
-  struct r_debug *r = _dl_debug_initialize (0, nsid);
+  struct r_debug *r_debug_compat;
+  struct r_debug *r = _dl_debug_initialize (0, nsid, &r_debug_compat);
   bool make_consistent = false;
 
   /* Get file information.  To match the kernel behavior, do not fill
@@ -986,6 +987,8 @@ _dl_map_object_from_fd (const char *name, const char *origname, int fd,
 	  if (make_consistent && r != NULL)
 	    {
 	      r->r_state = RT_CONSISTENT;
+	      if (r_debug_compat != NULL)
+		r_debug_compat->r_state = RT_CONSISTENT;
 	      _dl_debug_state ();
 	      LIBC_PROBE (map_failed, 2, nsid, r);
 	    }
@@ -1082,6 +1085,8 @@ _dl_map_object_from_fd (const char *name, const char *origname, int fd,
 	 call _dl_debug_initialize in a static program in case dynamic
 	 linking has not been used before.  */
       r->r_state = RT_ADD;
+      if (r_debug_compat != NULL)
+	r_debug_compat->r_state = RT_ADD;
       _dl_debug_state ();
       LIBC_PROBE (map_start, 2, nsid, r);
       make_consistent = true;
diff --git a/elf/dl-open.c b/elf/dl-open.c
index ec386626f9..cf4f381b3f 100644
--- a/elf/dl-open.c
+++ b/elf/dl-open.c
@@ -523,7 +523,7 @@ dl_open_worker (void *a)
 
   /* One might be tempted to assert that we are RT_CONSISTENT at this point, but that
      may not be true if this is a recursive call to dlopen.  */
-  _dl_debug_initialize (0, args->nsid);
+  _dl_debug_initialize (0, args->nsid, NULL);
 
   /* Load the named object.  */
   struct link_map *new;
@@ -574,7 +574,8 @@ dl_open_worker (void *a)
       if ((mode & RTLD_GLOBAL) && new->l_global == 0)
 	add_to_global_update (new);
 
-      assert (_dl_debug_initialize (0, args->nsid)->r_state == RT_CONSISTENT);
+      assert (_dl_debug_initialize (0, args->nsid, NULL)->r_state
+	      == RT_CONSISTENT);
 
       return;
     }
@@ -630,8 +631,12 @@ dl_open_worker (void *a)
 #endif
 
   /* Notify the debugger all new objects are now ready to go.  */
-  struct r_debug *r = _dl_debug_initialize (0, args->nsid);
+  struct r_debug *r_debug_compat;
+  struct r_debug *r = _dl_debug_initialize (0, args->nsid,
+					    &r_debug_compat);
   r->r_state = RT_CONSISTENT;
+  if (r_debug_compat != NULL)
+    r_debug_compat->r_state = RT_CONSISTENT;
   _dl_debug_state ();
   LIBC_PROBE (map_complete, 3, args->nsid, r, new);
 
@@ -830,7 +835,7 @@ no more namespaces available for dlmopen()"));
 	  ++GL(dl_nns);
 	}
 
-      _dl_debug_initialize (0, nsid)->r_state = RT_CONSISTENT;
+      _dl_debug_initialize (0, nsid, NULL)->r_state = RT_CONSISTENT;
     }
   /* Never allow loading a DSO in a namespace which is empty.  Such
      direct placements is only causing problems.  Also don't allow
@@ -899,7 +904,8 @@ no more namespaces available for dlmopen()"));
 	     the flag here.  */
 	}
 
-      assert (_dl_debug_initialize (0, args.nsid)->r_state == RT_CONSISTENT);
+      assert (_dl_debug_initialize (0, args.nsid, NULL)->r_state
+	      == RT_CONSISTENT);
 
       /* Release the lock.  */
       __rtld_lock_unlock_recursive (GL(dl_load_lock));
@@ -908,7 +914,8 @@ no more namespaces available for dlmopen()"));
       _dl_signal_exception (errcode, &exception, NULL);
     }
 
-  assert (_dl_debug_initialize (0, args.nsid)->r_state == RT_CONSISTENT);
+  assert (_dl_debug_initialize (0, args.nsid, NULL)->r_state
+	  == RT_CONSISTENT);
 
   /* Release the lock.  */
   __rtld_lock_unlock_recursive (GL(dl_load_lock));
diff --git a/elf/dl-reloc-static-pie.c b/elf/dl-reloc-static-pie.c
index d5bd2f31e9..30c8f1a8bc 100644
--- a/elf/dl-reloc-static-pie.c
+++ b/elf/dl-reloc-static-pie.c
@@ -52,7 +52,7 @@ _dl_relocate_static_pie (void)
   main_map->l_relocated = 1;
 
   /* Initialize _r_debug.  */
-  struct r_debug *r = _dl_debug_initialize (0, LM_ID_BASE);
+  struct r_debug *r = _dl_debug_initialize (0, LM_ID_BASE, NULL);
   r->r_state = RT_CONSISTENT;
 
   /* Set up debugging before the debugger is notified for the first
@@ -66,5 +66,11 @@ _dl_relocate_static_pie (void)
        with the run-time address of the r_debug structure  */
     main_map->l_info[DT_DEBUG]->d_un.d_ptr = (ElfW(Addr)) r;
 # endif
+
+  /* There is a DT_DEBUGSZ entry in the dynamic section.  Fill in the
+     size of the r_debug structure  */
+  if (main_map->l_info[VERSYMIDX (DT_DEBUGSZ)] != NULL)
+    main_map->l_info[VERSYMIDX (DT_DEBUGSZ)]->d_un.d_val
+      = sizeof (struct r_debug);
 }
 #endif
diff --git a/elf/elf.h b/elf/elf.h
index 4738dfa28f..f9f4d217c2 100644
--- a/elf/elf.h
+++ b/elf/elf.h
@@ -939,6 +939,7 @@ typedef struct
    GNU extension.  */
 #define DT_VERSYM	0x6ffffff0
 
+#define DT_DEBUGSZ	0x6ffffff8
 #define DT_RELACOUNT	0x6ffffff9
 #define DT_RELCOUNT	0x6ffffffa
 
diff --git a/elf/link.h b/elf/link.h
index ff3a85c847..229b22c57a 100644
--- a/elf/link.h
+++ b/elf/link.h
@@ -61,9 +61,20 @@ struct r_debug
       } r_state;
 
     ElfW(Addr) r_ldbase;	/* Base address the linker is loaded at.  */
+
+    /* The following extended fields should be accessed only after
+       verifying the size of the r_debug structure via the DT_DEBUGSZ
+       dynamic tag.  */
+
+    /* Link to next r_debug structure.  Each r_debug structure represents
+       a different namespace.  The first r_debug structure is for the
+       default namespace.  */
+    struct r_debug *next;
   };
 
-/* This is the instance of that structure used by the dynamic linker.  */
+/* This is the compatibility symbol of that structure provided by the
+   dynamic linker.  Access to its fields beyond r_ldbase may be invalid.
+ */
 extern struct r_debug _r_debug;
 
 /* This symbol refers to the "dynamic structure" in the `.dynamic' section
diff --git a/elf/rtld-debugger-interface.txt b/elf/rtld-debugger-interface.txt
index 61bc99e4b0..920da9c21a 100644
--- a/elf/rtld-debugger-interface.txt
+++ b/elf/rtld-debugger-interface.txt
@@ -32,6 +32,18 @@ but there is no way for the debugger to discover whether any of the
 objects in the link-map have been relocated or not.
 
 
+Extension to the r_debug structure
+==================================
+
+If the executable's dynamic section has a DT_DEBUGSZ element, the
+run-time linker sets that element's value to the size of the extended
+r_debug structure which has the additional field:
+
+  struct r_debug *next;
+    Link to next r_debug structure.  Each r_debug structure represents a
+    different namespace.  The first r_debug structure is for the default
+    namespace.
+
 Probe-based debugger interface
 ==============================
 
diff --git a/elf/rtld.c b/elf/rtld.c
index 878e6480f4..a652dd3729 100644
--- a/elf/rtld.c
+++ b/elf/rtld.c
@@ -1662,7 +1662,7 @@ dl_main (const ElfW(Phdr) *phdr,
 
   /* Initialize _r_debug.  */
   struct r_debug *r = _dl_debug_initialize (GL(dl_rtld_map).l_addr,
-					    LM_ID_BASE);
+					    LM_ID_BASE, NULL);
   r->r_state = RT_CONSISTENT;
 
   /* Put the link_map for ourselves on the chain so it can be found by
@@ -1771,6 +1771,12 @@ dl_main (const ElfW(Phdr) *phdr,
     GL(dl_rtld_map).l_info[DT_DEBUG]->d_un.d_ptr = (ElfW(Addr)) r;
 #endif
 
+  /* There is a DT_DEBUGSZ entry in the dynamic section.  Fill in the
+     size of the r_debug structure  */
+  if (main_map->l_info[VERSYMIDX (DT_DEBUGSZ)] != NULL)
+    main_map->l_info[VERSYMIDX (DT_DEBUGSZ)]->d_un.d_val
+      = sizeof (struct r_debug);
+
   /* We start adding objects.  */
   r->r_state = RT_ADD;
   _dl_debug_state ();
@@ -2491,7 +2497,7 @@ dl_main (const ElfW(Phdr) *phdr,
 
   /* Notify the debugger all new objects are now ready to go.  We must re-get
      the address since by now the variable might be in another object.  */
-  r = _dl_debug_initialize (0, LM_ID_BASE);
+  r = _dl_debug_initialize (0, LM_ID_BASE, NULL);
   r->r_state = RT_CONSISTENT;
   _dl_debug_state ();
   LIBC_PROBE (init_complete, 2, LM_ID_BASE, r);
diff --git a/elf/tst-_r_gnu_debug.c b/elf/tst-_r_gnu_debug.c
new file mode 100644
index 0000000000..cf0f99c9da
--- /dev/null
+++ b/elf/tst-_r_gnu_debug.c
@@ -0,0 +1,60 @@
+/* Test _r_gnu_debug via DT_DEBUG and DT_DEBUGSZ.
+   Copyright (C) 2021 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <config.h>
+#include <link.h>
+#include <stdlib.h>
+#include <support/check.h>
+#include <support/test-driver.h>
+
+#ifndef ELF_MACHINE_GET_R_DEBUG
+# define ELF_MACHINE_GET_R_DEBUG(d) \
+    (__extension__ ({ 					\
+      struct r_debug *debug;				\
+      if ((d)->d_tag == DT_DEBUG)			\
+	debug = (struct r_debug *) (d)->d_un.d_ptr;	\
+      else						\
+	debug = NULL;					\
+      debug; }))
+#endif
+
+static int
+do_test (void)
+{
+  int found = 0;
+  ElfW(Dyn) *d;
+  for (d = _DYNAMIC; d->d_tag != DT_NULL; ++d)
+    if (d->d_tag == DT_DEBUGSZ)
+      {
+	TEST_VERIFY_EXIT (d->d_un.d_val == sizeof (struct r_debug));
+	found++;
+      }
+    else
+      {
+	struct r_debug *debug = ELF_MACHINE_GET_R_DEBUG (d);
+	if (debug != NULL)
+	  {
+	    TEST_VERIFY_EXIT (debug->r_version == 1);
+	    found++;
+	  }
+      }
+  return (found == (HAVE_Z_DT_DEBUGSZ ? 2 : 1)
+	  ? EXIT_SUCCESS : EXIT_UNSUPPORTED);
+}
+
+#include <support/test-driver.c>
diff --git a/include/link.h b/include/link.h
index 4af16cb596..a762fccdc1 100644
--- a/include/link.h
+++ b/include/link.h
@@ -353,6 +353,10 @@ struct auditstate
 };
 
 
+/* This is the hidden instance of struct r_debug used by the dynamic
+   linker.  */
+extern struct r_debug _r_gnu_debug attribute_hidden;
+
 #if __ELF_NATIVE_CLASS == 32
 # define symbind symbind32
 #elif __ELF_NATIVE_CLASS == 64
diff --git a/sysdeps/generic/ldsodefs.h b/sysdeps/generic/ldsodefs.h
index 9c15259236..c02027b309 100644
--- a/sysdeps/generic/ldsodefs.h
+++ b/sysdeps/generic/ldsodefs.h
@@ -1095,8 +1095,10 @@ rtld_hidden_proto (_dl_debug_state)
 
 /* Initialize `struct r_debug' if it has not already been done.  The
    argument is the run-time load address of the dynamic linker, to be put
-   in the `r_ldbase' member.  Returns the address of the structure.  */
-extern struct r_debug *_dl_debug_initialize (ElfW(Addr) ldbase, Lmid_t ns)
+   in the `r_ldbase' member.  Returns the address of the structure and
+   the address of the compatibility symbol.  */
+extern struct r_debug *_dl_debug_initialize (ElfW(Addr) ldbase, Lmid_t ns,
+					     struct r_debug **)
      attribute_hidden;
 
 /* Initialize the basic data structure for the search paths.  SOURCE
-- 
2.31.1

Re: [PATCH] Extend struct r_debug to support multiple namespaces

[Daniel Walker via Libc-alpha]

On Sat, Aug 14, 2021 at 05:33:36PM -0700, H.J. Lu wrote:
> On Mon, Aug 09, 2021 at 10:16:53AM -0700, Daniel Walker wrote:
> > On Mon, Aug 09, 2021 at 07:32:26AM -0700, H.J. Lu wrote:
> > > We need a new DT_XXX to support dlmopen.  We have 2 choices:
> > > 
> > > 1. Similar to DT_DEBUG.  Linker will allocate a new DT_XXX and
> > > ld.so will fill it with the address of the new debug data structure for
> > > dlmopen.
> > > 2. Similar to DT_MIPS_RLD_MAP_REL/DT_MIPS_RLD_MAP.
> > > Linker will allocate a space for a pointer, a new DT_XXX and fill
> > > the DT_XXX entry with the address of the pointer.   ld.so will update
> > > the pointer with the address of the new debug data structure for
> > > dlmopen.
> > > 
> > > #1 is the most straightforward to implement.  #2 is compatible with
> > > the current MIPS implementation.
> > > 
> > > Does anyone have any preferences?
> > 
> > 
> > I have #1 fully implemented already.
> > 
> 
> Here is a slightly different approach with DT_DEBUGSZ.  The corresponding
> linker patch is at
> 
> https://sourceware.org/pipermail/binutils/2021-August/117706.html
> 
> H.J.

What advantages does this provide over have a different structure ?

Lets say we wanted to extend the structure further in the future , is that
possible in this implementation ?

Daniel

Re: [PATCH] Extend struct r_debug to support multiple namespaces

[H.J. Lu via Libc-alpha]

On Mon, Aug 16, 2021 at 9:20 AM Daniel Walker <danielwa@cisco.com> wrote:
>
> On Sat, Aug 14, 2021 at 05:33:36PM -0700, H.J. Lu wrote:
> > On Mon, Aug 09, 2021 at 10:16:53AM -0700, Daniel Walker wrote:
> > > On Mon, Aug 09, 2021 at 07:32:26AM -0700, H.J. Lu wrote:
> > > > We need a new DT_XXX to support dlmopen.  We have 2 choices:
> > > >
> > > > 1. Similar to DT_DEBUG.  Linker will allocate a new DT_XXX and
> > > > ld.so will fill it with the address of the new debug data structure for
> > > > dlmopen.
> > > > 2. Similar to DT_MIPS_RLD_MAP_REL/DT_MIPS_RLD_MAP.
> > > > Linker will allocate a space for a pointer, a new DT_XXX and fill
> > > > the DT_XXX entry with the address of the pointer.   ld.so will update
> > > > the pointer with the address of the new debug data structure for
> > > > dlmopen.
> > > >
> > > > #1 is the most straightforward to implement.  #2 is compatible with
> > > > the current MIPS implementation.
> > > >
> > > > Does anyone have any preferences?
> > >
> > >
> > > I have #1 fully implemented already.
> > >
> >
> > Here is a slightly different approach with DT_DEBUGSZ.  The corresponding
> > linker patch is at
> >
> > https://sourceware.org/pipermail/binutils/2021-August/117706.html
> >
> > H.J.
>
> What advantages does this provide over have a different structure ?
>
> Lets say we wanted to extend the structure further in the future , is that
> possible in this implementation ?

I submitted the v2 patch.

-- 
H.J.