Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Ian Murray]

Hi,

This was originally posted to the user list but it was suggested to me to post it here also. I am getting a failed compilation when I try to compile Xen 4.5.0 from the tarball and also from Git. I can compile RC4, although a bisect reveals that they pertinent difference is debug=y in Config.mk, which allows it to compile. That compiles without a problem and that is what I am running with. I have tried to build on two different servers with the same result, however both are running Ubuntu 12.04.5 and are similarly configured.

Steps I (would) use to build are simply...

./configure
make dist
make install


which results in...

...
gcc -O2 -fomit-frame-pointer -m64 -fno-strict-aliasing -std=gnu99 -Wall 
-Wstrict-prototypes -Wdeclaration-after-statement -Wno-unused-but-set-variable  
-DNDEBUG -I/root/xen450/xen-4.5.0/xen/include  
-I/root/xen450/xen-4.5.0/xen/include/asm-x86/mach-generic 
-I/root/xen450/xen-4.5.0/xen/include/asm-x86/mach-default -msoft-float 
-fno-stack-protector -fno-exceptions -Wnested-externs -DHAVE_GAS_VMX 
-DHAVE_GAS_EPT -DHAVE_GAS_FSGSBASE -mno-red-zone -mno-sse -fpic 
-fno-asynchronous-unwind-tables -DGCC_HAS_VISIBILITY_ATTRIBUTE -fno-builtin 
-fno-common -Werror -Wredundant-decls -Wno-pointer-arith -pipe -g -D__XEN__ 
-include /root/xen450/xen-4.5.0/xen/include/xen/config.h -nostdinc -DHAS_ACPI 
-DHAS_GDBSX -DHAS_PASSTHROUGH -DHAS_MEM_ACCESS -DHAS_MEM_PAGING 
-DHAS_MEM_SHARING -DHAS_PCI -DHAS_IOPORTS -DHAS_PDX -MMD -MF .hvm.o.d -c hvm.c 
-o hvm.o
hvm.c: In function ‘hvm_create_ioreq_server’:
hvm.c:487:18: error: ‘bufioreq_pfn’ may be used uninitialised in this function 
[-Werror=uninitialized]
hvm.c:718:30: note: ‘bufioreq_pfn’ was declared here
cc1: all warnings being treated as errors
make[5]: *** [hvm.o] Error 1

Any suggestions are welcome,

Thanks for reading,

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Paul Durrant]

> -----Original Message-----
> From: xen-devel-bounces@lists.xen.org [mailto:xen-devel-
> bounces@lists.xen.org] On Behalf Of Ian Murray
> Sent: 15 March 2015 22:59
> To: xen-devel@lists.xensource.com
> Subject: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may
> be used uninitialised in this function [-Werror=uninitialized]
> 
> Hi,
> 
> This was originally posted to the user list but it was suggested to me to post it
> here also. I am getting a failed compilation when I try to compile Xen 4.5.0
> from the tarball and also from Git. I can compile RC4, although a bisect
> reveals that they pertinent difference is debug=y in Config.mk, which allows
> it to compile. That compiles without a problem and that is what I am running
> with. I have tried to build on two different servers with the same result,
> however both are running Ubuntu 12.04.5 and are similarly configured.
> 
> Steps I (would) use to build are simply...
> 
> ./configure
> make dist
> make install
> 
> 
> which results in...
> 
> ...
> gcc -O2 -fomit-frame-pointer -m64 -fno-strict-aliasing -std=gnu99 -Wall
> -Wstrict-prototypes -Wdeclaration-after-statement -Wno-unused-but-set-
> variable
> -DNDEBUG -I/root/xen450/xen-4.5.0/xen/include
> -I/root/xen450/xen-4.5.0/xen/include/asm-x86/mach-generic
> -I/root/xen450/xen-4.5.0/xen/include/asm-x86/mach-default -msoft-float
> -fno-stack-protector -fno-exceptions -Wnested-externs -DHAVE_GAS_VMX
> -DHAVE_GAS_EPT -DHAVE_GAS_FSGSBASE -mno-red-zone -mno-sse -fpic
> -fno-asynchronous-unwind-tables -DGCC_HAS_VISIBILITY_ATTRIBUTE -fno-
> builtin
> -fno-common -Werror -Wredundant-decls -Wno-pointer-arith -pipe -g -
> D__XEN__
> -include /root/xen450/xen-4.5.0/xen/include/xen/config.h -nostdinc -
> DHAS_ACPI
> -DHAS_GDBSX -DHAS_PASSTHROUGH -DHAS_MEM_ACCESS -
> DHAS_MEM_PAGING
> -DHAS_MEM_SHARING -DHAS_PCI -DHAS_IOPORTS -DHAS_PDX -MMD -MF
> .hvm.o.d -c hvm.c
> -o hvm.o
> hvm.c: In function ‘hvm_create_ioreq_server’:
> hvm.c:487:18: error: ‘bufioreq_pfn’ may be used uninitialised in this function
> [-Werror=uninitialized]
> hvm.c:718:30: note: ‘bufioreq_pfn’ was declared here
> cc1: all warnings being treated as errors
> make[5]: *** [hvm.o] Error 1
> 
> Any suggestions are welcome,
> 

Those line numbers don't work for me. I did a checkout of RELEASE-4.5.0 and, whilst bufioreq_pfn is indeed declared on line 718, I see no reference to it on line 487. Also, if I compile debug=n I see no problem. Is it possible you don't have a clean checkout of 4.5.0? What version of gcc are you using?

  Paul

> Thanks for reading,
> 
> Ian.
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Ian Campbell]

On Mon, 2015-03-16 at 09:45 +0000, Paul Durrant wrote:
> Those line numbers don't work for me. I did a checkout of
> RELEASE-4.5.0 and, whilst bufioreq_pfn is indeed declared on line 718,
> I see no reference to it on line 487. Also, if I compile debug=n I see
> no problem. Is it possible you don't have a clean checkout of 4.5.0?
> What version of gcc are you using?

Doing
 git show RELEASE-4.5.0:xen/arch/x86/hvm/hvm.c  > x
gets me a file x where line 487 is the first line (defn of "i") of:

static void hvm_free_ioreq_gmfn(struct domain *d, unsigned long gmfn)
{
    unsigned int i = gmfn - d->arch.hvm_domain.ioreq_gmfn.base;

    clear_bit(i, &d->arch.hvm_domain.ioreq_gmfn.mask);
}

My guess was that it was the call from the fial3 label in
hvm_ioreq_server_map_pages(). I can see how gcc wouldn't always be able
to reason about it always being initialised.

Ian.

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Ian Murray]

----- Original Message -----
> From: Paul Durrant <Paul.Durrant@citrix.com>
> To: Ian Murray <murrayie@yahoo.co.uk>; "xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>
> Cc: 
> Sent: Monday, 16 March 2015, 9:45
> Subject: Re: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function  [-Werror=uninitialized]
> 
>>  -----Original Message-----
>>  From: xen-devel-bounces@lists.xen.org [mailto:xen-devel-
>>  bounces@lists.xen.org] On Behalf Of Ian Murray
>>  Sent: 15 March 2015 22:59
>>  To: xen-devel@lists.xensource.com
>>  Subject: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ 
> may
>>  be used uninitialised in this function [-Werror=uninitialized]
>> 
>>  Hi,
>> 
>>  This was originally posted to the user list but it was suggested to me to 
> post it
>>  here also. I am getting a failed compilation when I try to compile Xen 
> 4.5.0
>>  from the tarball and also from Git. I can compile RC4, although a bisect
>>  reveals that they pertinent difference is debug=y in Config.mk, which 
> allows
>>  it to compile. That compiles without a problem and that is what I am 
> running
>>  with. I have tried to build on two different servers with the same result,
>>  however both are running Ubuntu 12.04.5 and are similarly configured.
>> 
>>  Steps I (would) use to build are simply...
>> 
>>  ./configure
>>  make dist
>>  make install
>> 
>> 
>>  which results in...
>> 
>>  ...
>>  gcc -O2 -fomit-frame-pointer -m64 -fno-strict-aliasing -std=gnu99 -Wall
>>  -Wstrict-prototypes -Wdeclaration-after-statement -Wno-unused-but-set-
>>  variable
>>  -DNDEBUG -I/root/xen450/xen-4.5.0/xen/include
>>  -I/root/xen450/xen-4.5.0/xen/include/asm-x86/mach-generic
>>  -I/root/xen450/xen-4.5.0/xen/include/asm-x86/mach-default -msoft-float
>>  -fno-stack-protector -fno-exceptions -Wnested-externs -DHAVE_GAS_VMX
>>  -DHAVE_GAS_EPT -DHAVE_GAS_FSGSBASE -mno-red-zone -mno-sse -fpic
>>  -fno-asynchronous-unwind-tables -DGCC_HAS_VISIBILITY_ATTRIBUTE -fno-
>>  builtin
>>  -fno-common -Werror -Wredundant-decls -Wno-pointer-arith -pipe -g -
>>  D__XEN__
>>  -include /root/xen450/xen-4.5.0/xen/include/xen/config.h -nostdinc -
>>  DHAS_ACPI
>>  -DHAS_GDBSX -DHAS_PASSTHROUGH -DHAS_MEM_ACCESS -
>>  DHAS_MEM_PAGING
>>  -DHAS_MEM_SHARING -DHAS_PCI -DHAS_IOPORTS -DHAS_PDX -MMD -MF
>>  .hvm.o.d -c hvm.c
>>  -o hvm.o
>>  hvm.c: In function ‘hvm_create_ioreq_server’:
>>  hvm.c:487:18: error: ‘bufioreq_pfn’ may be used uninitialised in this 
> function
>>  [-Werror=uninitialized]
>>  hvm.c:718:30: note: ‘bufioreq_pfn’ was declared here
>>  cc1: all warnings being treated as errors
>>  make[5]: *** [hvm.o] Error 1
>> 
>>  Any suggestions are welcome,
>> 
> 
> Those line numbers don't work for me. I did a checkout of RELEASE-4.5.0 and, 
> whilst bufioreq_pfn is indeed declared on line 718, I see no reference to it on 
> line 487. Also, if I compile debug=n I see no problem. Is it possible you 
> don't have a clean checkout of 4.5.0? What version of gcc are you using?
> 
>   Paul
>

Thanks for replying.

# gcc --version
gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3

This is both from a brand new clone of Git and also the release tarball. Ian C has commented elsewhere about what the compiler might be up to, although it's beyond my knowledge in terms of how "clever" the compiler is being. FWIW, I couldn't really understand the line numbering, so I looked at the files themselves and couldn't see a direct, either.... and surely the variable in question is well out of scope at that point. (obviously I am being naive about something here.)





  
> 
>>  Thanks for reading,
>> 
>>  Ian.
>> 
>>  _______________________________________________
>>  Xen-devel mailing list
>>  Xen-devel@lists.xen.org
>>  http://lists.xen.org/xen-devel
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
>

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Slutz, Donald Christopher]

On 03/16/15 17:19, Ian Murray wrote:
> ----- Original Message -----
>> From: Paul Durrant <Paul.Durrant@citrix.com>
>> To: Ian Murray <murrayie@yahoo.co.uk>; "xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>
>> Cc: 
>> Sent: Monday, 16 March 2015, 9:45
>> Subject: Re: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function  [-Werror=uninitialized]
>>
>>>  -----Original Message-----
>>>  From: xen-devel-bounces@lists.xen.org [mailto:xen-devel-
>>>  bounces@lists.xen.org] On Behalf Of Ian Murray
>>>  Sent: 15 March 2015 22:59
>>>  To: xen-devel@lists.xensource.com
>>>  Subject: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ 
>> may
>>>  be used uninitialised in this function [-Werror=uninitialized]
>>>
...
>>>
>>>  Any suggestions are welcome,
>>>
>>
>> Those line numbers don't work for me. I did a checkout of RELEASE-4.5.0 and, 
>> whilst bufioreq_pfn is indeed declared on line 718, I see no reference to it on 
>> line 487. Also, if I compile debug=n I see no problem. Is it possible you 
>> don't have a clean checkout of 4.5.0? What version of gcc are you using?
>>
>>   Paul
>>
> 
> Thanks for replying.
> 
> # gcc --version
> gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
> 
> This is both from a brand new clone of Git and also the release
> tarball. Ian C has commented elsewhere about what the compiler
> might be up to, although it's beyond my knowledge in terms of
> how "clever" the compiler is being. FWIW, I couldn't really
> understand the line numbering, so I looked at the files
> themselves and couldn't see a direct, either.... and surely the
> variable in question is well out of scope at that
> point. (obviously I am being naive about something here.)
> 
> 
> 

The gcc I am using: gcc (GCC) 4.7.2 20120921 (Red Hat 4.7.2-2) reported
the same error (with adjusted line numbers) for code that I am working
on.  The reference is really:



fail3:
    if ( !is_default && handle_bufioreq )
        hvm_free_ioreq_gmfn(d, bufioreq_pfn);


So I had assumed that I had uncovered a gcc bug, since the only path
here to "hvm_free_ioreq_gmfn(d, bufioreq_pfn)" requires
"handle_bufioreq" to be true and not the goto for fail2.  It looks to me
like  bufioreq_pfn is always set if you get to fail3's call on
"hvm_free_ioreq_gmfn". This report looks to same to me.  I am not able
to see your issue, but I am planning on including:


diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 72be5b9..cb6c763 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -715,7 +715,7 @@ static int hvm_ioreq_server_map_pages(struct
hvm_ioreq_server *s,
                                       bool_t is_default, bool_t
handle_bufioreq)
 {
     struct domain *d = s->domain;
-    unsigned long ioreq_pfn, bufioreq_pfn;
+    unsigned long ioreq_pfn, bufioreq_pfn = 0;
     int rc;

     if ( is_default )


Which "fixed" it for me.  It would be good for you to try this.

   -Don Slutz

> 
> 
>   
>>
>>>  Thanks for reading,
>>>
>>>  Ian.
>>>
>>>  _______________________________________________
>>>  Xen-devel mailing list
>>>  Xen-devel@lists.xen.org
>>>  http://lists.xen.org/xen-devel
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xen.org
>> http://lists.xen.org/xen-devel
>>
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
> 
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Ian Murray]

On 17/03/15 00:10, Slutz, Donald Christopher wrote:
> On 03/16/15 17:19, Ian Murray wrote:
>> ----- Original Message -----
>>> From: Paul Durrant <Paul.Durrant@citrix.com>
>>> To: Ian Murray <murrayie@yahoo.co.uk>; "xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>
>>> Cc: 
>>> Sent: Monday, 16 March 2015, 9:45
>>> Subject: Re: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function  [-Werror=uninitialized]
>>>
>>>>  -----Original Message-----
>>>>  From: xen-devel-bounces@lists.xen.org [mailto:xen-devel-
>>>>  bounces@lists.xen.org] On Behalf Of Ian Murray
>>>>  Sent: 15 March 2015 22:59
>>>>  To: xen-devel@lists.xensource.com
>>>>  Subject: [Xen-devel] Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ 
>>> may
>>>>  be used uninitialised in this function [-Werror=uninitialized]
>>>>
> ...
>>>>  Any suggestions are welcome,
>>>>
>>> Those line numbers don't work for me. I did a checkout of RELEASE-4.5.0 and, 
>>> whilst bufioreq_pfn is indeed declared on line 718, I see no reference to it on 
>>> line 487. Also, if I compile debug=n I see no problem. Is it possible you 
>>> don't have a clean checkout of 4.5.0? What version of gcc are you using?
>>>
>>>   Paul
>>>
>> Thanks for replying.
>>
>> # gcc --version
>> gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
>>
>> This is both from a brand new clone of Git and also the release
>> tarball. Ian C has commented elsewhere about what the compiler
>> might be up to, although it's beyond my knowledge in terms of
>> how "clever" the compiler is being. FWIW, I couldn't really
>> understand the line numbering, so I looked at the files
>> themselves and couldn't see a direct, either.... and surely the
>> variable in question is well out of scope at that
>> point. (obviously I am being naive about something here.)
>>
>>
>>
> The gcc I am using: gcc (GCC) 4.7.2 20120921 (Red Hat 4.7.2-2) reported
> the same error (with adjusted line numbers) for code that I am working
> on.  The reference is really:
>
>
>
> fail3:
>     if ( !is_default && handle_bufioreq )
>         hvm_free_ioreq_gmfn(d, bufioreq_pfn);
>
>
> So I had assumed that I had uncovered a gcc bug, since the only path
> here to "hvm_free_ioreq_gmfn(d, bufioreq_pfn)" requires
> "handle_bufioreq" to be true and not the goto for fail2.  It looks to me
> like  bufioreq_pfn is always set if you get to fail3's call on
> "hvm_free_ioreq_gmfn". This report looks to same to me.  I am not able
> to see your issue, but I am planning on including:
>
>
> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> index 72be5b9..cb6c763 100644
> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -715,7 +715,7 @@ static int hvm_ioreq_server_map_pages(struct
> hvm_ioreq_server *s,
>                                        bool_t is_default, bool_t
> handle_bufioreq)
>  {
>      struct domain *d = s->domain;
> -    unsigned long ioreq_pfn, bufioreq_pfn;
> +    unsigned long ioreq_pfn, bufioreq_pfn = 0;
>      int rc;
>
>      if ( is_default )
>
>
> Which "fixed" it for me.  It would be good for you to try this.
>
>    -Don Slutz

Thanks for reply. Elsewhere Ian Campbell got to the same conclusion
regarding the real "issue" being around fail3. I tried zeroing the
variable earlier tonight... as much to test to see if there were further
issues down the line as anything. Xen compiled without error. Thanks again.




>>
>>   
>>>>  Thanks for reading,
>>>>
>>>>  Ian.
>>>>
>>>>  _______________________________________________
>>>>  Xen-devel mailing list
>>>>  Xen-devel@lists.xen.org
>>>>  http://lists.xen.org/xen-devel
>>> _______________________________________________
>>> Xen-devel mailing list
>>> Xen-devel@lists.xen.org
>>> http://lists.xen.org/xen-devel
>>>
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xen.org
>> http://lists.xen.org/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Ian Campbell]

On Tue, 2015-03-17 at 00:10 +0000, Slutz, Donald Christopher wrote:
> @@ -715,7 +715,7 @@ static int hvm_ioreq_server_map_pages(struct
> hvm_ioreq_server *s,
>                                        bool_t is_default, bool_t
> handle_bufioreq)
>  {
>      struct domain *d = s->domain;
> -    unsigned long ioreq_pfn, bufioreq_pfn;
> +    unsigned long ioreq_pfn, bufioreq_pfn = 0;
>      int rc;
> 
>      if ( is_default )
> 
> 
> Which "fixed" it for me.  It would be good for you to try this.

Are you going to send this as a proper patch? I think usually we would
add a brief comment to explain the unnecessary seeming initialiser. e..g
in radix-tree.c "/* uninitialised var warning */" or "/* gcc ... */" in
various other places.

Re: Compliling Xen 4.5.0 Fails with error: ‘bufioreq_pfn’ may be used uninitialised in this function [-Werror=uninitialized]

[Slutz, Donald Christopher]

On 03/17/15 06:37, Ian Campbell wrote:
> On Tue, 2015-03-17 at 00:10 +0000, Slutz, Donald Christopher wrote:
>> @@ -715,7 +715,7 @@ static int hvm_ioreq_server_map_pages(struct
>> hvm_ioreq_server *s,
>>                                        bool_t is_default, bool_t
>> handle_bufioreq)
>>  {
>>      struct domain *d = s->domain;
>> -    unsigned long ioreq_pfn, bufioreq_pfn;
>> +    unsigned long ioreq_pfn, bufioreq_pfn = 0;
>>      int rc;
>>
>>      if ( is_default )
>>
>>
>> Which "fixed" it for me.  It would be good for you to try this.
> 
> Are you going to send this as a proper patch? I think usually we would
> add a brief comment to explain the unnecessary seeming initialiser. e..g
> in radix-tree.c "/* uninitialised var warning */" or "/* gcc ... */" in
> various other places.
> 
> 

I do not yet have this as a separate patch.  Will do so.

   -Don Slutz