Am 5. Februar 2023 11:26:10 UTC schrieb "Philippe Mathieu-Daudé"
<phi...@linaro.org>:
>On 4/2/23 16:10, Bernhard Beschow wrote:
>> When setting up the CPU's smram memory region alias, the code currently
>> assumes that the smram size is 4 GiB. While this is true, it repeats a
>> decision made elsewhere which seems redundant and prone to
>> inconsistencies. Avoid this by reusing whatever size the smram region
>> was set to.
>>
>> Signed-off-by: Bernhard Beschow <shen...@gmail.com>
>> ---
>> target/i386/tcg/sysemu/tcg-cpu.c | 3 +--
>> 1 file changed, 1 insertion(+), 2 deletions(-)
>>
>> diff --git a/target/i386/tcg/sysemu/tcg-cpu.c
>> b/target/i386/tcg/sysemu/tcg-cpu.c
>> index c223c0fe9b..8f5ce6093c 100644
>> --- a/target/i386/tcg/sysemu/tcg-cpu.c
>> +++ b/target/i386/tcg/sysemu/tcg-cpu.c
>> @@ -22,7 +22,6 @@
>> #include "tcg/helper-tcg.h"
>> #include "sysemu/sysemu.h"
>> -#include "qemu/units.h"
>> #include "exec/address-spaces.h"
>> #include "tcg/tcg-cpu.h"
>> @@ -36,7 +35,7 @@ static void tcg_cpu_machine_done(Notifier *n, void *unused)
>> if (smram) {
>> cpu->smram = g_new(MemoryRegion, 1);
>> memory_region_init_alias(cpu->smram, OBJECT(cpu), "smram",
>> - smram, 0, 4 * GiB);
>> + smram, 0, memory_region_size(smram));
>
>Or define SMRAM_REGION_SIZE and use it?
This could still lead to inconsistencies if the size was changed in one place
only, no?
>
>(subject header can be shortened to "target/i386:").
Okay.