Re: [1/2] kernel32/tests: Fix compilation with PSDK.

[Michael Stefaniuc]

Hello Dmitry,

On 08/19/2015 06:23 AM, Dmitry Timoshkov wrote:
> dlls/kernel32/tests/process.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/dlls/kernel32/tests/process.c b/dlls/kernel32/tests/process.c
> index 113ec35..a8abdc1 100644
> --- a/dlls/kernel32/tests/process.c
> +++ b/dlls/kernel32/tests/process.c
> @@ -2302,7 +2302,7 @@ static void test_TerminateJobObject(void)
>
> static void test_QueryInformationJobObject(void)
>  {
> - char buf[FIELD_OFFSET(JOBOBJECT_BASIC_PROCESS_ID_LIST, ProcessIdList[5])];
> - char buf[sizeof(JOBOBJECT_BASIC_PROCESS_ID_LIST) + sizeof(ULONG_PTR) * 5];

this change looks odd.

Using offsetof (or FIELD_OFFSET) is the standard Wine way of getting the
size of a struct with a variable length array. I have done quite a few
patches on Alexandre's request that replace sizeof+sizeof*x with
offsetof. E.g. 14e20162f9448a648f3973a86e03aea17387c040

Last but not least you're not allocating the same amount of memory:
  offsetof(struct, field[n]) == sizeof(struct) + (n-1) * sizeof(field)

bye
    michael

[1/2] kernel32/tests: Fix compilation with PSDK.

[Dmitry Timoshkov]

Michael Stefaniuc  wrote:
> > - char buf[FIELD_OFFSET(JOBOBJECT_BASIC_PROCESS_ID_LIST, ProcessIdList[5])];
> > - char buf[sizeof(JOBOBJECT_BASIC_PROCESS_ID_LIST) + sizeof(ULONG_PTR) * 5];
>
> this change looks odd.
>
> Using offsetof (or FIELD_OFFSET) is the standard Wine way of getting the
> size of a struct with a variable length array. I have done quite a few
> patches on Alexandre's request that replace sizeof+sizeof*x with
> offsetof. E.g. 14e20162f9448a648f3973a86e03aea17387c040

Well, it was already discussed before, MSVC/PSDK compilers don't cope with
such a construct, FIELD_OFFSET() is not a constant for them, and they fail.
An ability to compile with MSVC/PSDK compilers is very critical thing at least
for writing Wine tests, so we have to take that limitation into account.
> Last but not least you're not allocating the same amount of memory:
>   offsetof(struct, field[n]) == sizeof(struct) + (n-1) * sizeof(field)

Yes, I decided to make it visible that 5 fields are explicitely allocated
instead of adding a comment that 1 field is already allocated in the base
structure, but it shouldn't matter.