Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Josh Poimboeuf]

On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
> From: Miroslav Benes <mbenes@suse.cz>
> 
> Josh reported a bug:
> 
>   When the object to be patched is a module, and that module is
>   rmmod'ed and reloaded, it fails to load with:
> 
>   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
>   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
>   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> 
>   The livepatch module has a relocation which references a symbol
>   in the _previous_ loading of nfsd. When apply_relocate_add()
>   tries to replace the old relocation with a new one, it sees that
>   the previous one is nonzero and it errors out.
> 
>   On ppc64le, we have a similar issue:
> 
>   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
>   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
>   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'

Shouldn't there also be a fix for this powerpc issue?

-- 
Josh

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Song Liu]

On Wed, Jan 18, 2023 at 2:08 PM Josh Poimboeuf <jpoimboe@kernel.org> wrote:
>
> On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
> > From: Miroslav Benes <mbenes@suse.cz>
> >
> > Josh reported a bug:
> >
> >   When the object to be patched is a module, and that module is
> >   rmmod'ed and reloaded, it fails to load with:
> >
> >   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
> >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> >
> >   The livepatch module has a relocation which references a symbol
> >   in the _previous_ loading of nfsd. When apply_relocate_add()
> >   tries to replace the old relocation with a new one, it sees that
> >   the previous one is nonzero and it errors out.
> >
> >   On ppc64le, we have a similar issue:
> >
> >   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
> >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
>
> Shouldn't there also be a fix for this powerpc issue?

There was a working version, but it was not very clean. We couldn't agree
on the path forward for powerpc, so we are hoping to ship the fix to x86 (and
s390?) first [1].

Thanks,
Song

[1] https://lore.kernel.org/live-patching/Y7hLvpHqgY0oJ4GY@alley/#t

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Josh Poimboeuf]

On Thu, Jan 19, 2023 at 11:06:35AM -0800, Song Liu wrote:
> On Wed, Jan 18, 2023 at 2:08 PM Josh Poimboeuf <jpoimboe@kernel.org> wrote:
> >
> > On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
> > > From: Miroslav Benes <mbenes@suse.cz>
> > >
> > > Josh reported a bug:
> > >
> > >   When the object to be patched is a module, and that module is
> > >   rmmod'ed and reloaded, it fails to load with:
> > >
> > >   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
> > >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> > >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> > >
> > >   The livepatch module has a relocation which references a symbol
> > >   in the _previous_ loading of nfsd. When apply_relocate_add()
> > >   tries to replace the old relocation with a new one, it sees that
> > >   the previous one is nonzero and it errors out.
> > >
> > >   On ppc64le, we have a similar issue:
> > >
> > >   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
> > >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> > >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> >
> > Shouldn't there also be a fix for this powerpc issue?
> 
> There was a working version, but it was not very clean. We couldn't agree
> on the path forward for powerpc, so we are hoping to ship the fix to x86 (and
> s390?) first [1].

Sorry for coming in late, I was on leave so I missed a lot of the
discussions on previous versions.  The decision to leave powerpc broken
wasn't clear from reading the commit message.  The bug is mentioned, and
the fix is implied, but surprisingly there's no fix.

I agree that the powerpc fix should be in a separate patch, but I still
don't feel comfortable merging the x86 fix without the corresponding
powerpc fix.

powerpc is a major arch and not a second-class citizen.  If we don't fix
it now then it'll probably never get fixed until it blows up in the real
world.

For powerpc, instead of clearing, how about just "fixing" the warning
site, something like so (untested)?


diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c
index 1096d6b3a62c..1a12463ba674 100644
--- a/arch/powerpc/kernel/module_64.c
+++ b/arch/powerpc/kernel/module_64.c
@@ -499,9 +499,11 @@ static unsigned long stub_for_addr(const Elf64_Shdr *sechdrs,
 
 /* We expect a noop next: if it is, replace it with instruction to
    restore r2. */
-static int restore_r2(const char *name, u32 *instruction, struct module *me)
+static int restore_r2(const char *name, u32 *instruction, struct module *me,
+		      bool klp_sym)
 {
 	u32 *prev_insn = instruction - 1;
+	u32 insn_val = *instruction;
 
 	if (is_mprofile_ftrace_call(name))
 		return 1;
@@ -514,9 +516,18 @@ static int restore_r2(const char *name, u32 *instruction, struct module *me)
 	if (!instr_is_relative_link_branch(ppc_inst(*prev_insn)))
 		return 1;
 
-	if (*instruction != PPC_RAW_NOP()) {
+	/*
+	 * For a livepatch relocation, the restore r2 instruction might have
+	 * been previously written if the relocation references a symbol in a
+	 * module which was unloaded and is now being reloaded.  In that case,
+	 * skip the warning and instruction write.
+	 */
+	if (klp_sym && insn_val == PPC_INST_LD_TOC)
+		return 0;
+
+	if (insn_val != PPC_RAW_NOP()) {
 		pr_err("%s: Expected nop after call, got %08x at %pS\n",
-			me->name, *instruction, instruction);
+			me->name, insn_val, instruction);
 		return 0;
 	}
 
@@ -649,7 +660,8 @@ int apply_relocate_add(Elf64_Shdr *sechdrs,
 				if (!value)
 					return -ENOENT;
 				if (!restore_r2(strtab + sym->st_name,
-							(u32 *)location + 1, me))
+						(u32 *)location + 1, me,
+						sym->st_shndx == SHN_LIVEPATCH))
 					return -ENOEXEC;
 			} else
 				value += local_entry_offset(sym);

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Song Liu]

On Thu, Jan 19, 2023 at 10:42 PM Josh Poimboeuf <jpoimboe@kernel.org> wrote:
>
> On Thu, Jan 19, 2023 at 11:06:35AM -0800, Song Liu wrote:
> > On Wed, Jan 18, 2023 at 2:08 PM Josh Poimboeuf <jpoimboe@kernel.org> wrote:
> > >
> > > On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
> > > > From: Miroslav Benes <mbenes@suse.cz>
> > > >
> > > > Josh reported a bug:
> > > >
> > > >   When the object to be patched is a module, and that module is
> > > >   rmmod'ed and reloaded, it fails to load with:
> > > >
> > > >   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
> > > >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> > > >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> > > >
> > > >   The livepatch module has a relocation which references a symbol
> > > >   in the _previous_ loading of nfsd. When apply_relocate_add()
> > > >   tries to replace the old relocation with a new one, it sees that
> > > >   the previous one is nonzero and it errors out.
> > > >
> > > >   On ppc64le, we have a similar issue:
> > > >
> > > >   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
> > > >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> > > >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> > >
> > > Shouldn't there also be a fix for this powerpc issue?
> >
> > There was a working version, but it was not very clean. We couldn't agree
> > on the path forward for powerpc, so we are hoping to ship the fix to x86 (and
> > s390?) first [1].
>
> Sorry for coming in late, I was on leave so I missed a lot of the
> discussions on previous versions.  The decision to leave powerpc broken
> wasn't clear from reading the commit message.  The bug is mentioned, and
> the fix is implied, but surprisingly there's no fix.
>
> I agree that the powerpc fix should be in a separate patch, but I still
> don't feel comfortable merging the x86 fix without the corresponding
> powerpc fix.
>
> powerpc is a major arch and not a second-class citizen.  If we don't fix
> it now then it'll probably never get fixed until it blows up in the real
> world.
>
> For powerpc, instead of clearing, how about just "fixing" the warning
> site, something like so (untested)?

This version looks reasonable to me.

Thanks,
Song

>
> diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c
> index 1096d6b3a62c..1a12463ba674 100644
> --- a/arch/powerpc/kernel/module_64.c
> +++ b/arch/powerpc/kernel/module_64.c
> @@ -499,9 +499,11 @@ static unsigned long stub_for_addr(const Elf64_Shdr *sechdrs,
>
>  /* We expect a noop next: if it is, replace it with instruction to
>     restore r2. */
> -static int restore_r2(const char *name, u32 *instruction, struct module *me)
> +static int restore_r2(const char *name, u32 *instruction, struct module *me,
> +                     bool klp_sym)
>  {
>         u32 *prev_insn = instruction - 1;
> +       u32 insn_val = *instruction;
>
>         if (is_mprofile_ftrace_call(name))
>                 return 1;
> @@ -514,9 +516,18 @@ static int restore_r2(const char *name, u32 *instruction, struct module *me)
>         if (!instr_is_relative_link_branch(ppc_inst(*prev_insn)))
>                 return 1;
>
> -       if (*instruction != PPC_RAW_NOP()) {
> +       /*
> +        * For a livepatch relocation, the restore r2 instruction might have
> +        * been previously written if the relocation references a symbol in a
> +        * module which was unloaded and is now being reloaded.  In that case,
> +        * skip the warning and instruction write.
> +        */
> +       if (klp_sym && insn_val == PPC_INST_LD_TOC)
> +               return 0;
> +
> +       if (insn_val != PPC_RAW_NOP()) {
>                 pr_err("%s: Expected nop after call, got %08x at %pS\n",
> -                       me->name, *instruction, instruction);
> +                       me->name, insn_val, instruction);
>                 return 0;
>         }
>
> @@ -649,7 +660,8 @@ int apply_relocate_add(Elf64_Shdr *sechdrs,
>                                 if (!value)
>                                         return -ENOENT;
>                                 if (!restore_r2(strtab + sym->st_name,
> -                                                       (u32 *)location + 1, me))
> +                                               (u32 *)location + 1, me,
> +                                               sym->st_shndx == SHN_LIVEPATCH))
>                                         return -ENOEXEC;
>                         } else
>                                 value += local_entry_offset(sym);

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Josh Poimboeuf]

On Fri, Jan 20, 2023 at 09:03:55AM -0800, Song Liu wrote:
> > > > Shouldn't there also be a fix for this powerpc issue?
> > >
> > > There was a working version, but it was not very clean. We couldn't agree
> > > on the path forward for powerpc, so we are hoping to ship the fix to x86 (and
> > > s390?) first [1].
> >
> > Sorry for coming in late, I was on leave so I missed a lot of the
> > discussions on previous versions.  The decision to leave powerpc broken
> > wasn't clear from reading the commit message.  The bug is mentioned, and
> > the fix is implied, but surprisingly there's no fix.
> >
> > I agree that the powerpc fix should be in a separate patch, but I still
> > don't feel comfortable merging the x86 fix without the corresponding
> > powerpc fix.
> >
> > powerpc is a major arch and not a second-class citizen.  If we don't fix
> > it now then it'll probably never get fixed until it blows up in the real
> > world.
> >
> > For powerpc, instead of clearing, how about just "fixing" the warning
> > site, something like so (untested)?
> 
> This version looks reasonable to me.

Ok, I'll run it through testing and work up a proper patch.  I just
noticed the one I posted has a major bug thanks to restore_r2()'s
surprising return semantics.

-- 
Josh

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Joe Lawrence]

On 1/20/23 01:42, Josh Poimboeuf wrote:
> On Thu, Jan 19, 2023 at 11:06:35AM -0800, Song Liu wrote:
>> On Wed, Jan 18, 2023 at 2:08 PM Josh Poimboeuf <jpoimboe@kernel.org> wrote:
>>>
>>> On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
>>>> From: Miroslav Benes <mbenes@suse.cz>
>>>>
>>>> Josh reported a bug:
>>>>
>>>>   When the object to be patched is a module, and that module is
>>>>   rmmod'ed and reloaded, it fails to load with:
>>>>
>>>>   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
>>>>   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
>>>>   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
>>>>
>>>>   The livepatch module has a relocation which references a symbol
>>>>   in the _previous_ loading of nfsd. When apply_relocate_add()
>>>>   tries to replace the old relocation with a new one, it sees that
>>>>   the previous one is nonzero and it errors out.
>>>>
>>>>   On ppc64le, we have a similar issue:
>>>>
>>>>   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
>>>>   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
>>>>   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
>>>
>>> Shouldn't there also be a fix for this powerpc issue?
>>
>> There was a working version, but it was not very clean. We couldn't agree
>> on the path forward for powerpc, so we are hoping to ship the fix to x86 (and
>> s390?) first [1].
> 
> Sorry for coming in late, I was on leave so I missed a lot of the
> discussions on previous versions.  The decision to leave powerpc broken
> wasn't clear from reading the commit message.  The bug is mentioned, and
> the fix is implied, but surprisingly there's no fix.
> 
> I agree that the powerpc fix should be in a separate patch, but I still
> don't feel comfortable merging the x86 fix without the corresponding
> powerpc fix.
> 
> powerpc is a major arch and not a second-class citizen.  If we don't fix
> it now then it'll probably never get fixed until it blows up in the real
> world.
> 
> For powerpc, instead of clearing, how about just "fixing" the warning
> site, something like so (untested)?
> 
> 
> diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c
> index 1096d6b3a62c..1a12463ba674 100644
> --- a/arch/powerpc/kernel/module_64.c
> +++ b/arch/powerpc/kernel/module_64.c
> @@ -499,9 +499,11 @@ static unsigned long stub_for_addr(const Elf64_Shdr *sechdrs,
>  
>  /* We expect a noop next: if it is, replace it with instruction to
>     restore r2. */
> -static int restore_r2(const char *name, u32 *instruction, struct module *me)
> +static int restore_r2(const char *name, u32 *instruction, struct module *me,
> +		      bool klp_sym)
>  {
>  	u32 *prev_insn = instruction - 1;
> +	u32 insn_val = *instruction;
>  
>  	if (is_mprofile_ftrace_call(name))
>  		return 1;
> @@ -514,9 +516,18 @@ static int restore_r2(const char *name, u32 *instruction, struct module *me)
>  	if (!instr_is_relative_link_branch(ppc_inst(*prev_insn)))
>  		return 1;
>  
> -	if (*instruction != PPC_RAW_NOP()) {
> +	/*
> +	 * For a livepatch relocation, the restore r2 instruction might have
> +	 * been previously written if the relocation references a symbol in a
> +	 * module which was unloaded and is now being reloaded.  In that case,
> +	 * skip the warning and instruction write.
> +	 */
> +	if (klp_sym && insn_val == PPC_INST_LD_TOC)
> +		return 0;

Hi Josh,

Nit: shouldn't this return 1?

And if you're willing to entertain a small refactor, wouldn't
restore_r2() be clearer if it returned -ESOMETHING on error?

Maybe converting to a boolean could work, but then I'd suggest a name
that clearly implies success/fail given true/false return.  Maybe
replace_nop_with_ld_toc() or replace_nop_to_restore_r2() ... still
-ESOMETHING is more intuitive to me as there are cases like this where
the function safely returns w/o replacing anything.

> +
> +	if (insn_val != PPC_RAW_NOP()) {
>  		pr_err("%s: Expected nop after call, got %08x at %pS\n",
> -			me->name, *instruction, instruction);
> +			me->name, insn_val, instruction);
>  		return 0;
>  	}
>  
> @@ -649,7 +660,8 @@ int apply_relocate_add(Elf64_Shdr *sechdrs,
>  				if (!value)
>  					return -ENOENT;
>  				if (!restore_r2(strtab + sym->st_name,
> -							(u32 *)location + 1, me))
> +						(u32 *)location + 1, me,
> +						sym->st_shndx == SHN_LIVEPATCH))
>  					return -ENOEXEC;
>  			} else
>  				value += local_entry_offset(sym);
> 

klp-convert-tree tests* ran OK with this patch (with the nit fixed) on
top of Song's v10.  LMK if you want me to push a branch with some or all
of these patches for further testing.

* I removed the tests that check for relocation clearing, only tested
module reloading

-- 
Joe

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Josh Poimboeuf]

On Tue, Jan 24, 2023 at 03:08:27PM -0500, Joe Lawrence wrote:
> > +	/*
> > +	 * For a livepatch relocation, the restore r2 instruction might have
> > +	 * been previously written if the relocation references a symbol in a
> > +	 * module which was unloaded and is now being reloaded.  In that case,
> > +	 * skip the warning and instruction write.
> > +	 */
> > +	if (klp_sym && insn_val == PPC_INST_LD_TOC)
> > +		return 0;
> 
> Hi Josh,
> 
> Nit: shouldn't this return 1?
> 
> And if you're willing to entertain a small refactor, wouldn't
> restore_r2() be clearer if it returned -ESOMETHING on error?
> 
> Maybe converting to a boolean could work, but then I'd suggest a name
> that clearly implies success/fail given true/false return.  Maybe
> replace_nop_with_ld_toc() or replace_nop_to_restore_r2() ... still
> -ESOMETHING is more intuitive to me as there are cases like this where
> the function safely returns w/o replacing anything.

Indeed, and I actually already discovered that and made such changes,
just need to get around to posting the patches.

-- 
Josh

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Josh Poimboeuf]

On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
> From: Miroslav Benes <mbenes@suse.cz>
> 
> Josh reported a bug:
> 
>   When the object to be patched is a module, and that module is
>   rmmod'ed and reloaded, it fails to load with:
> 
>   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
>   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
>   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> 
>   The livepatch module has a relocation which references a symbol
>   in the _previous_ loading of nfsd. When apply_relocate_add()
>   tries to replace the old relocation with a new one, it sees that
>   the previous one is nonzero and it errors out.

Should we add a selftest to make sure this problem doesn't come back?

>   On ppc64le, we have a similar issue:
> 
>   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
>   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
>   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'

Just to clarify my previous comment, the reference to the powerpc issue
should be removed because it'll be fixed in a separate patch.

> He also proposed three different solutions. We could remove the error
> check in apply_relocate_add() introduced by commit eda9cec4c9a1
> ("x86/module: Detect and skip invalid relocations"). However the check
> is useful for detecting corrupted modules.
> 
> We could also deny the patched modules to be removed. If it proved to be
> a major drawback for users, we could still implement a different
> approach. The solution would also complicate the existing code a lot.
> 
> We thus decided to reverse the relocation patching (clear all relocation
> targets on x86_64). The solution is not
> universal and is too much arch-specific, but it may prove to be simpler
> in the end.
> 
> Signed-off-by: Miroslav Benes <mbenes@suse.cz>
> Signed-off-by: Song Liu <song@kernel.org>
> Acked-by: Miroslav Benes <mbenes@suse.cz>
> Co-developed-by: Song Liu <song@kernel.org>
> Reported-by: Josh Poimboeuf <jpoimboe@redhat.com>

According to submitting-patches.rst the Co-developed-by is supposed to
be immediately before your Signed-off-by.

Also, other than the commit log, this patch is almost completely
unrecognizable compared to Miroslav's original patch.  Does it still
make sense for him to be listed as the author?

In the -tip tree they sometimes use an Originally-by tag, which might be
relevant here.

> -static int __apply_relocate_add(Elf64_Shdr *sechdrs,
> +static int __write_relocate_add(Elf64_Shdr *sechdrs,
>  		   const char *strtab,
>  		   unsigned int symindex,
>  		   unsigned int relsec,
>  		   struct module *me,
> -		   void *(*write)(void *dest, const void *src, size_t len))
> +		   void *(*write)(void *dest, const void *src, size_t len),
> +		   bool apply)
>  {
>  	unsigned int i;
>  	Elf64_Rela *rel = (void *)sechdrs[relsec].sh_addr;
>  	Elf64_Sym *sym;
>  	void *loc;
>  	u64 val;
> +	u64 zero = 0ULL;
>  
> -	DEBUGP("Applying relocate section %u to %u\n",
> +	DEBUGP("%s relocate section %u to %u\n",
> +	       (apply) ? "Applying" : "Clearing",

"(apply)" has unnecessary parentheses.

>  	       relsec, sechdrs[relsec].sh_info);
>  	for (i = 0; i < sechdrs[relsec].sh_size / sizeof(*rel); i++) {
> +		int write_size = 4;

We already know we're writing, I suggest just calling it 'size' to be
more consistent with kernel style.

Also, it can be an unsigned value like size_t.

Also, instead of initializing it here with a potentially bogus value
which may need to be overwritten for a 64-bit reloc, it's better to
explicitly set the size in each individual case below.  That's makes the
logic clearer and helps prevent future bugs if new 64-bit relocation
types are added later.

>  		case R_X86_64_PC32:
>  		case R_X86_64_PLT32:
> -			if (*(u32 *)loc != 0)
> -				goto invalid_relocation;
>  			val -= (u64)loc;
> -			write(loc, &val, 4);
>  #if 0
> -			if ((s64)val != *(s32 *)loc)
> +			if ((s64)val != *(s32 *)&val)
>  				goto overflow;
>  #endif

Why is the compiled-out code getting changed?  Is it actually fixing a
hypothetical bug?

This code really needs to be removed anyway, it's been dead for at least
15 years.

>  			break;
>  		case R_X86_64_PC64:
> -			if (*(u64 *)loc != 0)
> -				goto invalid_relocation;
>  			val -= (u64)loc;
> -			write(loc, &val, 8);
> +			write_size = 8;
>  			break;
>  		default:
>  			pr_err("%s: Unknown rela relocation: %llu\n",
>  			       me->name, ELF64_R_TYPE(rel[i].r_info));
>  			return -ENOEXEC;
>  		}
> +
> +		if (apply) {
> +			if (memcmp(loc, &zero, write_size)) {
> +				pr_err("x86/modules: Skipping invalid relocation target, existing value is nonzero for type %d, loc %p, val %Lx\n",

It's not just "skipping", it's erroring out completely.  Yes, it's is a
pre-existing error message but we might as well improve it while
touching it.

Maybe just remove the "Skipping", i.e. change "Skipping invalid ..." to
"Invalid ..." ?

> +				       (int)ELF64_R_TYPE(rel[i].r_info), loc, val);
> +				return -ENOEXEC;
> +			}
> +			write(loc, &val, write_size);
> +		} else {
> +			if (memcmp(loc, &val, write_size)) {
> +				pr_warn("x86/modules: Clearing invalid relocation target, existing value does not match expected value for type %d, loc %p, val %Lx\n",
> +					(int)ELF64_R_TYPE(rel[i].r_info), loc, val);
> +			}
> +			write(loc, &zero, write_size);

If the value doesn't match then something has gone badly wrong.  Why go
ahead with the clearing in that case?

> +#ifdef CONFIG_LIVEPATCH
> +
> +void clear_relocate_add(Elf64_Shdr *sechdrs,
> +			const char *strtab,
> +			unsigned int symindex,
> +			unsigned int relsec,
> +			struct module *me)
> +{
> +	write_relocate_add(sechdrs, strtab, symindex, relsec, me, false);
> +}
> +#endif

Superflous newline after the '#ifdef CONFIG_LIVEPATCH'.

> +
>  #endif
>  
>  int module_finalize(const Elf_Ehdr *hdr,
> diff --git a/include/linux/moduleloader.h b/include/linux/moduleloader.h
> index 7b4587a19189..0b54ec9856df 100644
> --- a/include/linux/moduleloader.h
> +++ b/include/linux/moduleloader.h
> @@ -75,6 +75,13 @@ int apply_relocate_add(Elf_Shdr *sechdrs,
>  		       unsigned int symindex,
>  		       unsigned int relsec,
>  		       struct module *mod);
> +#ifdef CONFIG_LIVEPATCH

This could use a comment explaining the purpose of this function:

/*
 * Some architectures (namely x86_64 and ppc64) perform sanity checks when
 * applying relocations.  If a patched module gets unloaded and then later
 * reloaded (and re-patched), klp re-applies relocations to the replacement
 * function(s).  Any leftover relocations from the previous loading of the
 * patched module might trigger the sanity checks.
 *
 * To prevent that, when unloading a patched module, clear out any relocations
 * that might trigger arch-specific sanity checks on a future module reload.
 */

> +void clear_relocate_add(Elf_Shdr *sechdrs,
> +		   const char *strtab,
> +		   unsigned int symindex,
> +		   unsigned int relsec,
> +		   struct module *me);
> +#endif


-- 
Josh

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Song Liu]

On Fri, Jan 20, 2023 at 11:16 AM Josh Poimboeuf <jpoimboe@kernel.org> wrote:
>
> On Wed, Jan 18, 2023 at 12:47:28PM -0800, Song Liu wrote:
> > From: Miroslav Benes <mbenes@suse.cz>
> >
> > Josh reported a bug:
> >
> >   When the object to be patched is a module, and that module is
> >   rmmod'ed and reloaded, it fails to load with:
> >
> >   module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
> >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> >
> >   The livepatch module has a relocation which references a symbol
> >   in the _previous_ loading of nfsd. When apply_relocate_add()
> >   tries to replace the old relocation with a new one, it sees that
> >   the previous one is nonzero and it errors out.
>
> Should we add a selftest to make sure this problem doesn't come back?

IIRC, a selftest for this issue is not easy without Joe's klp-convert work.
At the moment I use kpatch-build for testing.

>
> >   On ppc64le, we have a similar issue:
> >
> >   module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
> >   livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> >   livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
>
> Just to clarify my previous comment, the reference to the powerpc issue
> should be removed because it'll be fixed in a separate patch.

Will remove.

>
> > He also proposed three different solutions. We could remove the error
> > check in apply_relocate_add() introduced by commit eda9cec4c9a1
> > ("x86/module: Detect and skip invalid relocations"). However the check
> > is useful for detecting corrupted modules.
> >
> > We could also deny the patched modules to be removed. If it proved to be
> > a major drawback for users, we could still implement a different
> > approach. The solution would also complicate the existing code a lot.
> >
> > We thus decided to reverse the relocation patching (clear all relocation
> > targets on x86_64). The solution is not
> > universal and is too much arch-specific, but it may prove to be simpler
> > in the end.
> >
> > Signed-off-by: Miroslav Benes <mbenes@suse.cz>
> > Signed-off-by: Song Liu <song@kernel.org>
> > Acked-by: Miroslav Benes <mbenes@suse.cz>
> > Co-developed-by: Song Liu <song@kernel.org>
> > Reported-by: Josh Poimboeuf <jpoimboe@redhat.com>
>
> According to submitting-patches.rst the Co-developed-by is supposed to
> be immediately before your Signed-off-by.
>
> Also, other than the commit log, this patch is almost completely
> unrecognizable compared to Miroslav's original patch.  Does it still
> make sense for him to be listed as the author?
>
> In the -tip tree they sometimes use an Originally-by tag, which might be
> relevant here.

How about:

Signed-off-by: Song Liu <song@kernel.org>
Originally-by: Miroslav Benes <mbenes@suse.cz>
Acked-by: Miroslav Benes <mbenes@suse.cz>
Reported-by: Josh Poimboeuf <jpoimboe@redhat.com>

>
> > -static int __apply_relocate_add(Elf64_Shdr *sechdrs,
> > +static int __write_relocate_add(Elf64_Shdr *sechdrs,
> >                  const char *strtab,
> >                  unsigned int symindex,
> >                  unsigned int relsec,
> >                  struct module *me,
> > -                void *(*write)(void *dest, const void *src, size_t len))
> > +                void *(*write)(void *dest, const void *src, size_t len),
> > +                bool apply)
> >  {
> >       unsigned int i;
> >       Elf64_Rela *rel = (void *)sechdrs[relsec].sh_addr;
> >       Elf64_Sym *sym;
> >       void *loc;
> >       u64 val;
> > +     u64 zero = 0ULL;
> >
> > -     DEBUGP("Applying relocate section %u to %u\n",
> > +     DEBUGP("%s relocate section %u to %u\n",
> > +            (apply) ? "Applying" : "Clearing",
>
> "(apply)" has unnecessary parentheses.
>
> >              relsec, sechdrs[relsec].sh_info);
> >       for (i = 0; i < sechdrs[relsec].sh_size / sizeof(*rel); i++) {
> > +             int write_size = 4;
>
> We already know we're writing, I suggest just calling it 'size' to be
> more consistent with kernel style.
>
> Also, it can be an unsigned value like size_t.
>
> Also, instead of initializing it here with a potentially bogus value
> which may need to be overwritten for a 64-bit reloc, it's better to
> explicitly set the size in each individual case below.  That's makes the
> logic clearer and helps prevent future bugs if new 64-bit relocation
> types are added later.

Will fix in v10.

>
> >               case R_X86_64_PC32:
> >               case R_X86_64_PLT32:
> > -                     if (*(u32 *)loc != 0)
> > -                             goto invalid_relocation;
> >                       val -= (u64)loc;
> > -                     write(loc, &val, 4);
> >  #if 0
> > -                     if ((s64)val != *(s32 *)loc)
> > +                     if ((s64)val != *(s32 *)&val)
> >                               goto overflow;
> >  #endif
>
> Why is the compiled-out code getting changed?  Is it actually fixing a
> hypothetical bug?

I just changed them the same way as other cases (assuming we remove
the #if 0, of course).

>
> This code really needs to be removed anyway, it's been dead for at least
> 15 years.

Shall we remove it now? Within the same patch? Or with a preparation
patch?

>
> >                       break;
> >               case R_X86_64_PC64:
> > -                     if (*(u64 *)loc != 0)
> > -                             goto invalid_relocation;
> >                       val -= (u64)loc;
> > -                     write(loc, &val, 8);
> > +                     write_size = 8;
> >                       break;
> >               default:
> >                       pr_err("%s: Unknown rela relocation: %llu\n",
> >                              me->name, ELF64_R_TYPE(rel[i].r_info));
> >                       return -ENOEXEC;
> >               }
> > +
> > +             if (apply) {
> > +                     if (memcmp(loc, &zero, write_size)) {
> > +                             pr_err("x86/modules: Skipping invalid relocation target, existing value is nonzero for type %d, loc %p, val %Lx\n",
>
> It's not just "skipping", it's erroring out completely.  Yes, it's is a
> pre-existing error message but we might as well improve it while
> touching it.
>
> Maybe just remove the "Skipping", i.e. change "Skipping invalid ..." to
> "Invalid ..." ?

Sounds good to me.

>
> > +                                    (int)ELF64_R_TYPE(rel[i].r_info), loc, val);
> > +                             return -ENOEXEC;
> > +                     }
> > +                     write(loc, &val, write_size);
> > +             } else {
> > +                     if (memcmp(loc, &val, write_size)) {
> > +                             pr_warn("x86/modules: Clearing invalid relocation target, existing value does not match expected value for type %d, loc %p, val %Lx\n",
> > +                                     (int)ELF64_R_TYPE(rel[i].r_info), loc, val);
> > +                     }
> > +                     write(loc, &zero, write_size);
>
> If the value doesn't match then something has gone badly wrong.  Why go
> ahead with the clearing in that case?

We can pr_err() then return -ENOEXEC (?). But I guess we need to
handle the error case in:
  klp_cleanup_module_patches_limited()
  klp_module_coming()
  klp_module_going()
and all the functions that call klp_module_going().

This seems a big overkill to me...

Or do you mean we just skip the write()?

>
> > +#ifdef CONFIG_LIVEPATCH
> > +
> > +void clear_relocate_add(Elf64_Shdr *sechdrs,
> > +                     const char *strtab,
> > +                     unsigned int symindex,
> > +                     unsigned int relsec,
> > +                     struct module *me)
> > +{
> > +     write_relocate_add(sechdrs, strtab, symindex, relsec, me, false);
> > +}
> > +#endif
>
> Superflous newline after the '#ifdef CONFIG_LIVEPATCH'.
>
> > +
> >  #endif
> >
> >  int module_finalize(const Elf_Ehdr *hdr,
> > diff --git a/include/linux/moduleloader.h b/include/linux/moduleloader.h
> > index 7b4587a19189..0b54ec9856df 100644
> > --- a/include/linux/moduleloader.h
> > +++ b/include/linux/moduleloader.h
> > @@ -75,6 +75,13 @@ int apply_relocate_add(Elf_Shdr *sechdrs,
> >                      unsigned int symindex,
> >                      unsigned int relsec,
> >                      struct module *mod);
> > +#ifdef CONFIG_LIVEPATCH
>
> This could use a comment explaining the purpose of this function:
>
> /*
>  * Some architectures (namely x86_64 and ppc64) perform sanity checks when
>  * applying relocations.  If a patched module gets unloaded and then later
>  * reloaded (and re-patched), klp re-applies relocations to the replacement
>  * function(s).  Any leftover relocations from the previous loading of the
>  * patched module might trigger the sanity checks.
>  *
>  * To prevent that, when unloading a patched module, clear out any relocations
>  * that might trigger arch-specific sanity checks on a future module reload.
>  */

Sounds great. Will add this to the next version.

>
> > +void clear_relocate_add(Elf_Shdr *sechdrs,
> > +                const char *strtab,
> > +                unsigned int symindex,
> > +                unsigned int relsec,
> > +                struct module *me);
> > +#endif
>
>
> --
> Josh

Re: [PATCH v9] livepatch: Clear relocation targets on a module removal

[Josh Poimboeuf]

On Fri, Jan 20, 2023 at 11:41:02AM -0800, Song Liu wrote:
> > >   The livepatch module has a relocation which references a symbol
> > >   in the _previous_ loading of nfsd. When apply_relocate_add()
> > >   tries to replace the old relocation with a new one, it sees that
> > >   the previous one is nonzero and it errors out.
> >
> > Should we add a selftest to make sure this problem doesn't come back?
> 
> IIRC, a selftest for this issue is not easy without Joe's klp-convert work.
> At the moment I use kpatch-build for testing.

Ah right, I remember that now.

> How about:
> 
> Signed-off-by: Song Liu <song@kernel.org>
> Originally-by: Miroslav Benes <mbenes@suse.cz>
> Acked-by: Miroslav Benes <mbenes@suse.cz>
> Reported-by: Josh Poimboeuf <jpoimboe@redhat.com>

Yes, but the ordering looks off, I think it should be more like:

Reported-by: Josh Poimboeuf <jpoimboe@redhat.com>
Originally-by: Miroslav Benes <mbenes@suse.cz>
Signed-off-by: Song Liu <song@kernel.org>
Acked-by: Miroslav Benes <mbenes@suse.cz>

And then make sure 'From:' is you.

BTW, this patch affects both livepatch and x86, so the subject prefix
should have "x86" added, something like:

  livepatch,x86: Clear relocations on module removal

> > This code really needs to be removed anyway, it's been dead for at least
> > 15 years.
> 
> Shall we remove it now? Within the same patch? Or with a preparation
> patch?
> 

A preparatory patch sounds good.

> > > +                                    (int)ELF64_R_TYPE(rel[i].r_info), loc, val);
> > > +                             return -ENOEXEC;
> > > +                     }
> > > +                     write(loc, &val, write_size);
> > > +             } else {
> > > +                     if (memcmp(loc, &val, write_size)) {
> > > +                             pr_warn("x86/modules: Clearing invalid relocation target, existing value does not match expected value for type %d, loc %p, val %Lx\n",
> > > +                                     (int)ELF64_R_TYPE(rel[i].r_info), loc, val);
> > > +                     }
> > > +                     write(loc, &zero, write_size);
> >
> > If the value doesn't match then something has gone badly wrong.  Why go
> > ahead with the clearing in that case?
> 
> We can pr_err() then return -ENOEXEC (?). But I guess we need to
> handle the error case in:
>   klp_cleanup_module_patches_limited()
>   klp_module_coming()
>   klp_module_going()
> and all the functions that call klp_module_going().
> 
> This seems a big overkill to me...
> 
> Or do you mean we just skip the write()?

At the very least, skip the write.

But I really think it should just break out of the loop and return an
error, there's no point in trying to continue clearing the rest of the
relocations if one of them failed.

It's probably fine for the callers to ignore the error, the module's
going to get unloaded regardless.

-- 
Josh