Re: git: 831b1ff7913f - main - UFS/FFS: Migrate to modern uintXX_t from u_intXX_t.

On Jul 27, 2023, at 17:22, Mark Millard <marklmi@yahoo.com> wrote:

> Kirk McKusick <mckusick_at_FreeBSD.org> wrote on
> Date: Thu, 27 Jul 2023 22:27:49 UTC :
> 
>> The branch main has been updated by mckusick:
>> 
>> URL: https://cgit.FreeBSD.org/src/commit/?id=831b1ff7913fb0b317a258a43dce5abefee03a31
>> 
>> commit 831b1ff7913fb0b317a258a43dce5abefee03a31
>> Author: Kirk McKusick <mckusick@FreeBSD.org>
>> AuthorDate: 2023-07-27 22:26:01 +0000
>> Commit: Kirk McKusick <mckusick@FreeBSD.org>
>> CommitDate: 2023-07-27 22:27:36 +0000
>> 
>> UFS/FFS: Migrate to modern uintXX_t from u_intXX_t.
>> 
>> As per https://lists.freebsd.org/archives/freebsd-scsi/2023-July/000257.html
>> move to the modern uintXX_t. While here also migrate u_char to uint8_t.
>> Where other kernel interfaces allow, migrate u_long to uint64_t.
> 
> Looks dangerous to me . . .
> 
> u_long (a.k.a. unsigned long) on armv7, for example, is
> 32 bits, not 64. Same for i386, I expect.

I forgot to mention 32-bit powerpc and could have just
quoted:

           LP64         ILP32 counterpart
           amd64        i386
           powerpc64    powerpc
           aarch64      armv6/armv7

and/or:

     On all supported architectures:

            Type         Size
            short        2
            int          4
            long         sizeof(void*)
            long long    8
            float        4
            double       8

> For reference, from /usr/include/sys/types.h :
> 
> #if __BSD_VISIBLE
> typedef unsigned char   u_char;
> typedef unsigned short  u_short;
> typedef unsigned int    u_int;
> typedef unsigned long   u_long;
> #ifndef _KERNEL
> typedef unsigned short  ushort;         /* Sys V compatibility */
> typedef unsigned int    uint;           /* Sys V compatibility */
> #endif
> #endif
> 
> Quoting FreeBSD's man arch:
> 
>     ILP32     int, long, void * types machine representations all have 4-byte
>               size.
> 
>     LP64      int type machine representation uses 4 bytes, while long and
>               void * are 8 bytes.
> 
> armv7 is ILP32, as, likely, i386 would be as well.
> 
> Note that u_int is 32 bits for both ILP32 and LP64, not 64.
> But there are a  lot of u_int -> uint64_t replacements in
> the update.
> 
>> No functional changes intended.
> 
> It would appear that it would take a bunch of validation to know if
> the size changes actually make no actual difference on the
> architectures where some sizes changed.
> 
>> MFC-after: 1 week
>> Sponsored-by: The FreeBSD Foundation
> . . .
> 

I also noticed an int vs. uint64_t switch (but the signed status does
not change the sizeof(...) results here vs. u_int / unsigned int):

-	len = roundup2(howmany(fs->fs_ncg, NBBY), sizeof(int));
+	len = roundup2(howmany(fs->fs_ncg, NBBY), sizeof(uint64_t));

===
Mark Millard
marklmi at yahoo.com