1. Patchew
  2. QEMU
  3. [Qemu-devel] [PATCH 0/2] Fix tests on recent gcc
  4. View patch

[Qemu-devel] [PATCH 2/2] tests: Make acpid test compile

Juan Quintela posted 2 patches 8 years, 5 months ago
There is a newer version of this series
[Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Juan Quintela 8 years, 5 months ago 
Compiler gets confused with the size of the struct, so move form
g_new0() to g_malloc0().

I *think* that the problem is in gcc (or glib for that matter), but
the documentation of the g_new0 states that 1sts first argument is an
struct type, and uint32_t is not an struct type.

Signed-off-by: Juan Quintela <quintela@redhat.com>
---
 tests/vmgenid-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/vmgenid-test.c b/tests/vmgenid-test.c
index 3d5c1c3615..032e1d465a 100644
--- a/tests/vmgenid-test.c
+++ b/tests/vmgenid-test.c
@@ -67,7 +67,7 @@ static uint32_t acpi_find_vgia(void)
     g_assert_cmpint(tables_nr, >, 0);
 
     /* get the addresses of the tables pointed by rsdt */
-    tables = g_new0(uint32_t, tables_nr);
+    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
     ACPI_READ_ARRAY_PTR(tables, tables_nr, rsdt);
 
     for (i = 0; i < tables_nr; i++) {
-- 
2.13.5
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Dr. David Alan Gilbert 8 years, 5 months ago 
* Juan Quintela (quintela@redhat.com) wrote:
> Compiler gets confused with the size of the struct, so move form
> g_new0() to g_malloc0().
> 
> I *think* that the problem is in gcc (or glib for that matter), but
> the documentation of the g_new0 states that 1sts first argument is an
> struct type, and uint32_t is not an struct type.
> 
> Signed-off-by: Juan Quintela <quintela@redhat.com>
> ---
>  tests/vmgenid-test.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/tests/vmgenid-test.c b/tests/vmgenid-test.c
> index 3d5c1c3615..032e1d465a 100644
> --- a/tests/vmgenid-test.c
> +++ b/tests/vmgenid-test.c
> @@ -67,7 +67,7 @@ static uint32_t acpi_find_vgia(void)
>      g_assert_cmpint(tables_nr, >, 0);
>  
>      /* get the addresses of the tables pointed by rsdt */
> -    tables = g_new0(uint32_t, tables_nr);
> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);

I think there's an easier fix for this I think;
try:

-    g_assert_cmpint(tables_nr, >, 0);
+    g_assert(tables_nr > 0);

Dave

>      ACPI_READ_ARRAY_PTR(tables, tables_nr, rsdt);
>  
>      for (i = 0; i < tables_nr; i++) {
> -- 
> 2.13.5
> 
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Cédric Le Goater 8 years, 5 months ago 
On 08/23/2017 01:53 PM, Dr. David Alan Gilbert wrote:
> * Juan Quintela (quintela@redhat.com) wrote:
>> Compiler gets confused with the size of the struct, so move form
>> g_new0() to g_malloc0().
>>
>> I *think* that the problem is in gcc (or glib for that matter), but
>> the documentation of the g_new0 states that 1sts first argument is an
>> struct type, and uint32_t is not an struct type.
>>
>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>> ---
>>  tests/vmgenid-test.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/tests/vmgenid-test.c b/tests/vmgenid-test.c
>> index 3d5c1c3615..032e1d465a 100644
>> --- a/tests/vmgenid-test.c
>> +++ b/tests/vmgenid-test.c
>> @@ -67,7 +67,7 @@ static uint32_t acpi_find_vgia(void)
>>      g_assert_cmpint(tables_nr, >, 0);
>>  
>>      /* get the addresses of the tables pointed by rsdt */
>> -    tables = g_new0(uint32_t, tables_nr);
>> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
> 
> I think there's an easier fix for this I think;
> try:
> 
> -    g_assert_cmpint(tables_nr, >, 0);
> +    g_assert(tables_nr > 0);

I fixed that one with :

@@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
     AcpiRsdpDescriptor rsdp_table;
     uint32_t rsdt;
     AcpiRsdtDescriptorRev1 rsdt_table;
-    int tables_nr;
+    uint32_t tables_nr;
     uint32_t *tables;
     AcpiTableHeader ssdt_table;
     VgidTable vgid_table;


C. 



> Dave
> 
>>      ACPI_READ_ARRAY_PTR(tables, tables_nr, rsdt);
>>  
>>      for (i = 0; i < tables_nr; i++) {
>> -- 
>> 2.13.5
>>
> --
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
>
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Eric Blake 8 years, 5 months ago 
On 08/28/2017 09:41 AM, Cédric Le Goater wrote:
> On 08/23/2017 01:53 PM, Dr. David Alan Gilbert wrote:
>> * Juan Quintela (quintela@redhat.com) wrote:
>>> Compiler gets confused with the size of the struct, so move form
>>> g_new0() to g_malloc0().
>>>
>>> I *think* that the problem is in gcc (or glib for that matter), but
>>> the documentation of the g_new0 states that 1sts first argument is an
>>> struct type, and uint32_t is not an struct type.
>>>
>>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>>> ---

>>>  
>>>      /* get the addresses of the tables pointed by rsdt */
>>> -    tables = g_new0(uint32_t, tables_nr);
>>> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
>>

> I fixed that one with :
> 
> @@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
>      AcpiRsdpDescriptor rsdp_table;
>      uint32_t rsdt;
>      AcpiRsdtDescriptorRev1 rsdt_table;
> -    int tables_nr;
> +    uint32_t tables_nr;

I like this one better (multiplication in g_malloc0() makes me worry
about overflow; using unsigned math to avoid the problem is nicer).  Are
we going to see a v2 of this patch series?

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Daniel P. Berrange 8 years, 5 months ago 
On Tue, Aug 29, 2017 at 03:17:25PM -0500, Eric Blake wrote:
> On 08/28/2017 09:41 AM, Cédric Le Goater wrote:
> > On 08/23/2017 01:53 PM, Dr. David Alan Gilbert wrote:
> >> * Juan Quintela (quintela@redhat.com) wrote:
> >>> Compiler gets confused with the size of the struct, so move form
> >>> g_new0() to g_malloc0().
> >>>
> >>> I *think* that the problem is in gcc (or glib for that matter), but
> >>> the documentation of the g_new0 states that 1sts first argument is an
> >>> struct type, and uint32_t is not an struct type.
> >>>
> >>> Signed-off-by: Juan Quintela <quintela@redhat.com>
> >>> ---
> 
> >>>  
> >>>      /* get the addresses of the tables pointed by rsdt */
> >>> -    tables = g_new0(uint32_t, tables_nr);
> >>> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
> >>
> 
> > I fixed that one with :
> > 
> > @@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
> >      AcpiRsdpDescriptor rsdp_table;
> >      uint32_t rsdt;
> >      AcpiRsdtDescriptorRev1 rsdt_table;
> > -    int tables_nr;
> > +    uint32_t tables_nr;
> 
> I like this one better (multiplication in g_malloc0() makes me worry
> about overflow; using unsigned math to avoid the problem is nicer).  Are
> we going to see a v2 of this patch series?

It should really be size_t, because it is assigned from the result of
a size_t calculation, but you then also need to change a later assert
which was relying on it being signed:

@@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
     AcpiRsdpDescriptor rsdp_table;
     uint32_t rsdt;
     AcpiRsdtDescriptorRev1 rsdt_table;
-    int tables_nr;
+    size_t tables_nr;
     uint32_t *tables;
     AcpiTableHeader ssdt_table;
     VgidTable vgid_table;
@@ -62,9 +62,9 @@ static uint32_t acpi_find_vgia(void)
     ACPI_ASSERT_CMP(rsdt_table.signature, "RSDT");
 
     /* compute the table entries in rsdt */
+    g_assert_cmpint(rsdt_table.length, >, sizeof(AcpiRsdtDescriptorRev1));
     tables_nr = (rsdt_table.length - sizeof(AcpiRsdtDescriptorRev1)) /
                 sizeof(uint32_t);
-    g_assert_cmpint(tables_nr, >, 0);
 
     /* get the addresses of the tables pointed by rsdt */
     tables = g_new0(uint32_t, tables_nr);


Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Juan Quintela 8 years, 5 months ago 
"Daniel P. Berrange" <berrange@redhat.com> wrote:
> On Tue, Aug 29, 2017 at 03:17:25PM -0500, Eric Blake wrote:
>> On 08/28/2017 09:41 AM, Cédric Le Goater wrote:
>> > On 08/23/2017 01:53 PM, Dr. David Alan Gilbert wrote:
>> >> * Juan Quintela (quintela@redhat.com) wrote:
>> >>> Compiler gets confused with the size of the struct, so move form
>> >>> g_new0() to g_malloc0().
>> >>>
>> >>> I *think* that the problem is in gcc (or glib for that matter), but
>> >>> the documentation of the g_new0 states that 1sts first argument is an
>> >>> struct type, and uint32_t is not an struct type.
>> >>>
>> >>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>> >>> ---
>> 
>> >>>  
>> >>>      /* get the addresses of the tables pointed by rsdt */
>> >>> -    tables = g_new0(uint32_t, tables_nr);
>> >>> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
>> >>
>> 
>> > I fixed that one with :
>> > 
>> > @@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
>> >      AcpiRsdpDescriptor rsdp_table;
>> >      uint32_t rsdt;
>> >      AcpiRsdtDescriptorRev1 rsdt_table;
>> > -    int tables_nr;
>> > +    uint32_t tables_nr;
>> 
>> I like this one better (multiplication in g_malloc0() makes me worry
>> about overflow; using unsigned math to avoid the problem is nicer).  Are
>> we going to see a v2 of this patch series?
>
> It should really be size_t, because it is assigned from the result of
> a size_t calculation, but you then also need to change a later assert
> which was relying on it being signed:
>
> @@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
>      AcpiRsdpDescriptor rsdp_table;
>      uint32_t rsdt;
>      AcpiRsdtDescriptorRev1 rsdt_table;
> -    int tables_nr;
> +    size_t tables_nr;

I was using this already.

>      uint32_t *tables;
>      AcpiTableHeader ssdt_table;
>      VgidTable vgid_table;
> @@ -62,9 +62,9 @@ static uint32_t acpi_find_vgia(void)
>      ACPI_ASSERT_CMP(rsdt_table.signature, "RSDT");
>  
>      /* compute the table entries in rsdt */
> +    g_assert_cmpint(rsdt_table.length, >, sizeof(AcpiRsdtDescriptorRev1));
>      tables_nr = (rsdt_table.length - sizeof(AcpiRsdtDescriptorRev1)) /
>                  sizeof(uint32_t);
> -    g_assert_cmpint(tables_nr, >, 0);
>  
>      /* get the addresses of the tables pointed by rsdt */
>      tables = g_new0(uint32_t, tables_nr);
>

And here we are, this mail arrived after I sent my new series.  Will
wait for more comments and resend later.

Thanks, Juan.
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Juan Quintela 8 years, 5 months ago 
Cedric Le Goater <clg@kaod.org> wrote:
> On 08/23/2017 01:53 PM, Dr. David Alan Gilbert wrote:
>> * Juan Quintela (quintela@redhat.com) wrote:
>>> Compiler gets confused with the size of the struct, so move form
>>> g_new0() to g_malloc0().
>>>
>>> I *think* that the problem is in gcc (or glib for that matter), but
>>> the documentation of the g_new0 states that 1sts first argument is an
>>> struct type, and uint32_t is not an struct type.
>>>
>>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>>> ---
>>>  tests/vmgenid-test.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/tests/vmgenid-test.c b/tests/vmgenid-test.c
>>> index 3d5c1c3615..032e1d465a 100644
>>> --- a/tests/vmgenid-test.c
>>> +++ b/tests/vmgenid-test.c
>>> @@ -67,7 +67,7 @@ static uint32_t acpi_find_vgia(void)
>>>      g_assert_cmpint(tables_nr, >, 0);
>>>  
>>>      /* get the addresses of the tables pointed by rsdt */
>>> -    tables = g_new0(uint32_t, tables_nr);
>>> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
>> 
>> I think there's an easier fix for this I think;
>> try:
>> 
>> -    g_assert_cmpint(tables_nr, >, 0);
>> +    g_assert(tables_nr > 0);

I liked more the following one.


>
> I fixed that one with :
>
> @@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
>      AcpiRsdpDescriptor rsdp_table;
>      uint32_t rsdt;
>      AcpiRsdtDescriptorRev1 rsdt_table;
> -    int tables_nr;
> +    uint32_t tables_nr;
>      uint32_t *tables;
>      AcpiTableHeader ssdt_table;
>      VgidTable vgid_table;


This make things work for me, so moving to this one.

Thanks, Juan.
Re: [Qemu-devel] [PATCH 2/2] tests: Make acpid test compile
Posted by Daniel P. Berrange 8 years, 5 months ago 
On Wed, Aug 30, 2017 at 12:51:23PM +0200, Juan Quintela wrote:
> Cedric Le Goater <clg@kaod.org> wrote:
> > On 08/23/2017 01:53 PM, Dr. David Alan Gilbert wrote:
> >> * Juan Quintela (quintela@redhat.com) wrote:
> >>> Compiler gets confused with the size of the struct, so move form
> >>> g_new0() to g_malloc0().
> >>>
> >>> I *think* that the problem is in gcc (or glib for that matter), but
> >>> the documentation of the g_new0 states that 1sts first argument is an
> >>> struct type, and uint32_t is not an struct type.
> >>>
> >>> Signed-off-by: Juan Quintela <quintela@redhat.com>
> >>> ---
> >>>  tests/vmgenid-test.c | 2 +-
> >>>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>>
> >>> diff --git a/tests/vmgenid-test.c b/tests/vmgenid-test.c
> >>> index 3d5c1c3615..032e1d465a 100644
> >>> --- a/tests/vmgenid-test.c
> >>> +++ b/tests/vmgenid-test.c
> >>> @@ -67,7 +67,7 @@ static uint32_t acpi_find_vgia(void)
> >>>      g_assert_cmpint(tables_nr, >, 0);
> >>>  
> >>>      /* get the addresses of the tables pointed by rsdt */
> >>> -    tables = g_new0(uint32_t, tables_nr);
> >>> +    tables = g_malloc0(sizeof(uint32_t) * tables_nr);
> >> 
> >> I think there's an easier fix for this I think;
> >> try:
> >> 
> >> -    g_assert_cmpint(tables_nr, >, 0);
> >> +    g_assert(tables_nr > 0);
> 
> I liked more the following one.
> 
> 
> >
> > I fixed that one with :
> >
> > @@ -40,7 +40,7 @@ static uint32_t acpi_find_vgia(void)
> >      AcpiRsdpDescriptor rsdp_table;
> >      uint32_t rsdt;
> >      AcpiRsdtDescriptorRev1 rsdt_table;
> > -    int tables_nr;
> > +    uint32_t tables_nr;
> >      uint32_t *tables;
> >      AcpiTableHeader ssdt_table;
> >      VgidTable vgid_table;
> 
> 
> This make things work for me, so moving to this one.

It should be size_t, because its taking the result of a calculation that
is size_t.  You also need to change the assert I mention in my other email.


Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.