diff mbox series

[RFC,PATCH-for-7.0] hw/i386/amd_iommu: Fix maybe-uninitialized error with GCC 12

Message ID 20220318004153.4510-1-philippe.mathieu.daude@gmail.com (mailing list archive)
State New, archived
Headers show
Series [RFC,PATCH-for-7.0] hw/i386/amd_iommu: Fix maybe-uninitialized error with GCC 12 | expand

Commit Message

Philippe Mathieu-Daudé March 18, 2022, 12:41 a.m. UTC
From: Philippe Mathieu-Daudé <f4bug@amsat.org>

Initialize 'oldlevel' early to avoid on Debian/sid:

  FAILED: libqemu-x86_64-softmmu.fa.p/hw_i386_amd_iommu.c.o
  In function 'pte_get_page_mask',
      inlined from 'amdvi_page_walk' at hw/i386/amd_iommu.c:945:25,
      inlined from 'amdvi_do_translate' at hw/i386/amd_iommu.c:989:5,
      inlined from 'amdvi_translate' at hw/i386/amd_iommu.c:1038:5:
  hw/i386/amd_iommu.c:877:38: error: 'oldlevel' may be used uninitialized [-Werror=maybe-uninitialized]
    877 |     return ~((1UL << ((oldlevel * 9) + 3)) - 1);
        |                      ~~~~~~~~~~~~~~~~^~~~
  hw/i386/amd_iommu.c: In function 'amdvi_translate':
  hw/i386/amd_iommu.c:906:41: note: 'oldlevel' was declared here
    906 |     unsigned level, present, pte_perms, oldlevel;
        |                                         ^~~~~~~~
  cc1: all warnings being treated as errors

Having:

  $ gcc --version
  gcc (Debian 12-20220313-1) 12.0.1 20220314 (experimental)

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
---
 hw/i386/amd_iommu.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Philippe Mathieu-Daudé March 18, 2022, 2:03 p.m. UTC | #1
+qemu-trivial@

On 18/3/22 01:41, Philippe Mathieu-Daudé wrote:
> From: Philippe Mathieu-Daudé <f4bug@amsat.org>
> 
> Initialize 'oldlevel' early to avoid on Debian/sid:
> 
>    FAILED: libqemu-x86_64-softmmu.fa.p/hw_i386_amd_iommu.c.o
>    In function 'pte_get_page_mask',
>        inlined from 'amdvi_page_walk' at hw/i386/amd_iommu.c:945:25,
>        inlined from 'amdvi_do_translate' at hw/i386/amd_iommu.c:989:5,
>        inlined from 'amdvi_translate' at hw/i386/amd_iommu.c:1038:5:
>    hw/i386/amd_iommu.c:877:38: error: 'oldlevel' may be used uninitialized [-Werror=maybe-uninitialized]
>      877 |     return ~((1UL << ((oldlevel * 9) + 3)) - 1);
>          |                      ~~~~~~~~~~~~~~~~^~~~
>    hw/i386/amd_iommu.c: In function 'amdvi_translate':
>    hw/i386/amd_iommu.c:906:41: note: 'oldlevel' was declared here
>      906 |     unsigned level, present, pte_perms, oldlevel;
>          |                                         ^~~~~~~~
>    cc1: all warnings being treated as errors
> 
> Having:
> 
>    $ gcc --version
>    gcc (Debian 12-20220313-1) 12.0.1 20220314 (experimental)
> 
> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> ---
>   hw/i386/amd_iommu.c | 1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/hw/i386/amd_iommu.c b/hw/i386/amd_iommu.c
> index 4d13d8e697..b6d299f964 100644
> --- a/hw/i386/amd_iommu.c
> +++ b/hw/i386/amd_iommu.c
> @@ -904,6 +904,7 @@ static void amdvi_page_walk(AMDVIAddressSpace *as, uint64_t *dte,
>       /* make sure the DTE has TV = 1 */
>       if (pte & AMDVI_DEV_TRANSLATION_VALID) {
>           level = get_pte_translation_mode(pte);
> +        oldlevel = level;
>           if (level >= 7) {
>               trace_amdvi_mode_invalid(level, addr);
>               return;
Paolo Bonzini March 21, 2022, 2:40 p.m. UTC | #2
On 3/18/22 15:03, Philippe Mathieu-Daudé wrote:
> +qemu-trivial@

Laurent,

please do not merge this patch as I've sent a replacement and I'll take 
care of merging it.

Paolo

> On 18/3/22 01:41, Philippe Mathieu-Daudé wrote:
>> From: Philippe Mathieu-Daudé <f4bug@amsat.org>
>>
>> Initialize 'oldlevel' early to avoid on Debian/sid:
>>
>>    FAILED: libqemu-x86_64-softmmu.fa.p/hw_i386_amd_iommu.c.o
>>    In function 'pte_get_page_mask',
>>        inlined from 'amdvi_page_walk' at hw/i386/amd_iommu.c:945:25,
>>        inlined from 'amdvi_do_translate' at hw/i386/amd_iommu.c:989:5,
>>        inlined from 'amdvi_translate' at hw/i386/amd_iommu.c:1038:5:
>>    hw/i386/amd_iommu.c:877:38: error: 'oldlevel' may be used 
>> uninitialized [-Werror=maybe-uninitialized]
>>      877 |     return ~((1UL << ((oldlevel * 9) + 3)) - 1);
>>          |                      ~~~~~~~~~~~~~~~~^~~~
>>    hw/i386/amd_iommu.c: In function 'amdvi_translate':
>>    hw/i386/amd_iommu.c:906:41: note: 'oldlevel' was declared here
>>      906 |     unsigned level, present, pte_perms, oldlevel;
>>          |                                         ^~~~~~~~
>>    cc1: all warnings being treated as errors
>>
>> Having:
>>
>>    $ gcc --version
>>    gcc (Debian 12-20220313-1) 12.0.1 20220314 (experimental)
>>
>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> ---
>>   hw/i386/amd_iommu.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/hw/i386/amd_iommu.c b/hw/i386/amd_iommu.c
>> index 4d13d8e697..b6d299f964 100644
>> --- a/hw/i386/amd_iommu.c
>> +++ b/hw/i386/amd_iommu.c
>> @@ -904,6 +904,7 @@ static void amdvi_page_walk(AMDVIAddressSpace *as, 
>> uint64_t *dte,
>>       /* make sure the DTE has TV = 1 */
>>       if (pte & AMDVI_DEV_TRANSLATION_VALID) {
>>           level = get_pte_translation_mode(pte);
>> +        oldlevel = level;
>>           if (level >= 7) {
>>               trace_amdvi_mode_invalid(level, addr);
>>               return;
> 
>
diff mbox series

Patch

diff --git a/hw/i386/amd_iommu.c b/hw/i386/amd_iommu.c
index 4d13d8e697..b6d299f964 100644
--- a/hw/i386/amd_iommu.c
+++ b/hw/i386/amd_iommu.c
@@ -904,6 +904,7 @@  static void amdvi_page_walk(AMDVIAddressSpace *as, uint64_t *dte,
     /* make sure the DTE has TV = 1 */
     if (pte & AMDVI_DEV_TRANSLATION_VALID) {
         level = get_pte_translation_mode(pte);
+        oldlevel = level;
         if (level >= 7) {
             trace_amdvi_mode_invalid(level, addr);
             return;