[Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Nikunj A Dadhania]

While the configure script generates TARGET_SUPPORTS_MTTCG define, one
of the define is cpus.c is checking wrong name: TARGET_SUPPORT_MTTCG

Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
---
 cpus.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/cpus.c b/cpus.c
index 68fdbc4..58d90aa 100644
--- a/cpus.c
+++ b/cpus.c
@@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
             } else if (use_icount) {
                 error_setg(errp, "No MTTCG when icount is enabled");
             } else {
-#ifndef TARGET_SUPPORT_MTTCG
+#ifndef TARGET_SUPPORTS_MTTCG
                 error_report("Guest not yet converted to MTTCG - "
                              "you may get unexpected results");
 #endif
-- 
2.9.3

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Alex Bennée]

Nikunj A Dadhania <nikunj@linux.vnet.ibm.com> writes:

> While the configure script generates TARGET_SUPPORTS_MTTCG define, one
> of the define is cpus.c is checking wrong name: TARGET_SUPPORT_MTTCG
>
> Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
> ---
>  cpus.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/cpus.c b/cpus.c
> index 68fdbc4..58d90aa 100644
> --- a/cpus.c
> +++ b/cpus.c
> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
>              } else if (use_icount) {
>                  error_setg(errp, "No MTTCG when icount is enabled");
>              } else {
> -#ifndef TARGET_SUPPORT_MTTCG
> +#ifndef TARGET_SUPPORTS_MTTCG
>                  error_report("Guest not yet converted to MTTCG - "
>                               "you may get unexpected results");
>  #endif

Opps that's embarrassing.

Applied to my tree, I'll be sending a pullreq today

--
Alex Bennée

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Richard Henderson]

On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:
> While the configure script generates TARGET_SUPPORTS_MTTCG define, one
> of the define is cpus.c is checking wrong name: TARGET_SUPPORT_MTTCG
>
> Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
> ---
>  cpus.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/cpus.c b/cpus.c
> index 68fdbc4..58d90aa 100644
> --- a/cpus.c
> +++ b/cpus.c
> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
>              } else if (use_icount) {
>                  error_setg(errp, "No MTTCG when icount is enabled");
>              } else {
> -#ifndef TARGET_SUPPORT_MTTCG
> +#ifndef TARGET_SUPPORTS_MTTCG

This sort of thing is why glibc moved to using -Wundef.

It would be a huge amount of work to convert our existing sources, but it would 
probably pay off in the long run.


r~

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Peter Maydell]

On 10 April 2017 at 18:26, Richard Henderson <rth@twiddle.net> wrote:
> On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:
>> --- a/cpus.c
>> +++ b/cpus.c
>> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
>>              } else if (use_icount) {
>>                  error_setg(errp, "No MTTCG when icount is enabled");
>>              } else {
>> -#ifndef TARGET_SUPPORT_MTTCG
>> +#ifndef TARGET_SUPPORTS_MTTCG
>
>
> This sort of thing is why glibc moved to using -Wundef.
>
> It would be a huge amount of work to convert our existing sources, but it
> would probably pay off in the long run.

We already build with -Wundef...

thanks
-- PMM

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Richard Henderson]

On 04/10/2017 10:29 AM, Peter Maydell wrote:
> On 10 April 2017 at 18:26, Richard Henderson <rth@twiddle.net> wrote:
>> On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:
>>> --- a/cpus.c
>>> +++ b/cpus.c
>>> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
>>>              } else if (use_icount) {
>>>                  error_setg(errp, "No MTTCG when icount is enabled");
>>>              } else {
>>> -#ifndef TARGET_SUPPORT_MTTCG
>>> +#ifndef TARGET_SUPPORTS_MTTCG
>>
>>
>> This sort of thing is why glibc moved to using -Wundef.
>>
>> It would be a huge amount of work to convert our existing sources, but it
>> would probably pay off in the long run.
>
> We already build with -Wundef...

Ah, but then you also have to stop using #ifdef and #ifndef, and use only #if. 
Thus my incomplete comment above about conversion.


r~

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Greg Kurz]

[-- Attachment #1: Type: text/plain, Size: 1042 bytes --]

On Mon, 10 Apr 2017 18:29:57 +0100
Peter Maydell <peter.maydell@linaro.org> wrote:

> On 10 April 2017 at 18:26, Richard Henderson <rth@twiddle.net> wrote:
> > On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:  
> >> --- a/cpus.c
> >> +++ b/cpus.c
> >> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
> >>              } else if (use_icount) {
> >>                  error_setg(errp, "No MTTCG when icount is enabled");
> >>              } else {
> >> -#ifndef TARGET_SUPPORT_MTTCG
> >> +#ifndef TARGET_SUPPORTS_MTTCG  
> >
> >
> > This sort of thing is why glibc moved to using -Wundef.
> >
> > It would be a huge amount of work to convert our existing sources, but it
> > would probably pay off in the long run.  
> 
> We already build with -Wundef...
> 

From the gcc info page:

'-Wundef'
     Warn if an undefined identifier is evaluated in an '#if' directive.

and BTW, isn't the purpose of #ifndef precisely to detect that the
identifier is undefined ?

> thanks
> -- PMM
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Richard Henderson]

On 04/10/2017 10:44 AM, Greg Kurz wrote:
> On Mon, 10 Apr 2017 18:29:57 +0100
> Peter Maydell <peter.maydell@linaro.org> wrote:
>
>> On 10 April 2017 at 18:26, Richard Henderson <rth@twiddle.net> wrote:
>>> On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:
>>>> --- a/cpus.c
>>>> +++ b/cpus.c
>>>> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
>>>>              } else if (use_icount) {
>>>>                  error_setg(errp, "No MTTCG when icount is enabled");
>>>>              } else {
>>>> -#ifndef TARGET_SUPPORT_MTTCG
>>>> +#ifndef TARGET_SUPPORTS_MTTCG
>>>
>>>
>>> This sort of thing is why glibc moved to using -Wundef.
>>>
>>> It would be a huge amount of work to convert our existing sources, but it
>>> would probably pay off in the long run.
>>
>> We already build with -Wundef...
>>
>
> From the gcc info page:
>
> '-Wundef'
>      Warn if an undefined identifier is evaluated in an '#if' directive.
>
> and BTW, isn't the purpose of #ifndef precisely to detect that the
> identifier is undefined ?

Yes, but it also has the typo problem above, whereas

   #if !TARGET_SUPPORTS_MTTCG

plus -Wundef would have caught that error.


r~

Re: [Qemu-devel] [PATCH fix for-2.9] cpus: fix wrong define name

[Greg Kurz]

[-- Attachment #1: Type: text/plain, Size: 1539 bytes --]

On Mon, 10 Apr 2017 10:49:12 -0700
Richard Henderson <rth@twiddle.net> wrote:

> On 04/10/2017 10:44 AM, Greg Kurz wrote:
> > On Mon, 10 Apr 2017 18:29:57 +0100
> > Peter Maydell <peter.maydell@linaro.org> wrote:
> >  
> >> On 10 April 2017 at 18:26, Richard Henderson <rth@twiddle.net> wrote:  
> >>> On 04/09/2017 11:06 PM, Nikunj A Dadhania wrote:  
> >>>> --- a/cpus.c
> >>>> +++ b/cpus.c
> >>>> @@ -202,7 +202,7 @@ void qemu_tcg_configure(QemuOpts *opts, Error **errp)
> >>>>              } else if (use_icount) {
> >>>>                  error_setg(errp, "No MTTCG when icount is enabled");
> >>>>              } else {
> >>>> -#ifndef TARGET_SUPPORT_MTTCG
> >>>> +#ifndef TARGET_SUPPORTS_MTTCG  
> >>>
> >>>
> >>> This sort of thing is why glibc moved to using -Wundef.
> >>>
> >>> It would be a huge amount of work to convert our existing sources, but it
> >>> would probably pay off in the long run.  
> >>
> >> We already build with -Wundef...
> >>  
> >
> > From the gcc info page:
> >
> > '-Wundef'
> >      Warn if an undefined identifier is evaluated in an '#if' directive.
> >
> > and BTW, isn't the purpose of #ifndef precisely to detect that the
> > identifier is undefined ?  
> 
> Yes, but it also has the typo problem above, whereas
> 
>    #if !TARGET_SUPPORTS_MTTCG
> 
> plus -Wundef would have caught that error.
> 

Turning such #ifdef/#ifndef to #if/#if ! would indeed make sense, but
we'd still have to be careful that every new addition follows the rule.

> 
> r~


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]