Re: [Patch v4 2/2] linux/bits.h: Add compile time sanity check of GENMASK inputs

[Rikard Falkeborn]

On Fri, Oct 11, 2019 at 07:27:37PM -0700, Andrew Morton wrote:
> On Wed,  9 Oct 2019 23:45:02 +0200 Rikard Falkeborn <rikard.falkeborn@xxxxxxxxx> wrote:
> 
> > GENMASK() and GENMASK_ULL() are supposed to be called with the high bit
> > as the first argument and the low bit as the second argument. Mixing
> > them will return a mask with zero bits set.
> > 
> > Recent commits show getting this wrong is not uncommon, see e.g.
> > commit aa4c0c9091b0 ("net: stmmac: Fix misuses of GENMASK macro") and
> > commit 9bdd7bb3a844 ("clocksource/drivers/npcm: Fix misuse of GENMASK
> > macro").
> > 
> > To prevent such mistakes from appearing again, add compile time sanity
> > checking to the arguments of GENMASK() and GENMASK_ULL(). If both
> > arguments are known at compile time, and the low bit is higher than the
> > high bit, break the build to detect the mistake immediately.
> > 
> > Since GENMASK() is used in declarations, BUILD_BUG_ON_ZERO() must be
> > used instead of BUILD_BUG_ON().
> > 
> > __builtin_constant_p does not evaluate is argument, it only checks if it
> > is a constant or not at compile time, and __builtin_choose_expr does not
> > evaluate the expression that is not chosen. Therefore, GENMASK(x++, 0)
> > does only evaluate x++ once.
> > 
> > Commit 95b980d62d52 ("linux/bits.h: make BIT(), GENMASK(), and friends
> > available in assembly") made the macros in linux/bits.h available in
> > assembly. Since BUILD_BUG_OR_ZERO() is not asm compatible, disable the
> > checks if the file is included in an asm file.
> > 
> > Due to bugs in GCC versions before 4.9 [0], disable the check if
> > building with a too old GCC compiler.
> > 
> > [0]: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=19449
> 
> I'm seeing some breakage in code which is newly added in linux-next:
> 
> sound/soc/codecs/tas2562.c: In function tas2562_set_dai_tdm_slot:
> ./include/linux/build_bug.h:16:51: error: negative width in bit-field <anonymous>
>  #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
>                                                    ^
> ./include/linux/bits.h:25:3: note: in expansion of macro BUILD_BUG_ON_ZERO
>   (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
>    ^~~~~~~~~~~~~~~~~
> ./include/linux/bits.h:39:3: note: in expansion of macro GENMASK_INPUT_CHECK
>   (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
>    ^~~~~~~~~~~~~~~~~~~
> sound/soc/codecs/tas2562.h:65:37: note: in expansion of macro GENMASK
>  #define TAS2562_TDM_CFG2_RXLEN_MASK GENMASK(0, 1)
>                                      ^~~~~~~
> sound/soc/codecs/tas2562.c:160:11: note: in expansion of macro TAS2562_TDM_CFG2_RXLEN_MASK
>            TAS2562_TDM_CFG2_RXLEN_MASK,
>            ^~~~~~~~~~~~~~~~~~~~~~~~~~~
> ./include/linux/build_bug.h:16:51: error: negative width in bit-field <anonymous>
>  #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
>                                                    ^
> ./include/linux/bits.h:25:3: note: in expansion of macro BUILD_BUG_ON_ZERO
>   (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
> 
> 
> and
> 
> In file included from ./include/linux/bits.h:23:0,
>                  from ./include/linux/ioport.h:15,
>                  from ./include/linux/acpi.h:12,
>                  from drivers/crypto/hisilicon/hpre/hpre_main.c:3:
> ./include/linux/build_bug.h:16:51: error: negative width in bit-field â<anonymous>â
>  #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
>                                                    ^
> ./include/linux/bits.h:25:3: note: in expansion of macro âBUILD_BUG_ON_ZEROâ
>   (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
>    ^~~~~~~~~~~~~~~~~
> ./include/linux/bits.h:39:3: note: in expansion of macro âGENMASK_INPUT_CHECKâ
>   (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
>    ^~~~~~~~~~~~~~~~~~~
> drivers/crypto/hisilicon/hpre/hpre_main.c:119:15: note: in expansion of macro âGENMASKâ
>   { .int_msk = GENMASK(10, 15), .msg = "hpre_ooo_rdrsp_err" },
>                ^~~~~~~
> ./include/linux/build_bug.h:16:51: error: negative width in bit-field â<anonymous>â
>  #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
>                                                    ^
> ./include/linux/bits.h:25:3: note: in expansion of macro âBUILD_BUG_ON_ZEROâ
>   (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
>    ^~~~~~~~~~~~~~~~~
> ./include/linux/bits.h:39:3: note: in expansion of macro âGENMASK_INPUT_CHECKâ
>   (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
>    ^~~~~~~~~~~~~~~~~~~
> drivers/crypto/hisilicon/hpre/hpre_main.c:120:15: note: in expansion of macro âGENMASKâ
>   { .int_msk = GENMASK(16, 21), .msg = "hpre_ooo_wrrsp_err" },
>                ^~~~~~~
> 

As of next-20191021, those breakages are fixed ([0], [1]). next-20191021
now builds (at least for x86-64) with the GENMASK compile checks added.

I also stumbled upon [2], which is a fix for an unused macro in a header
in mips/include/asm. next-201008 contained both v3 of this patch and the
mips header with the bad GENMASK in without build regressions, so it
*should* be ok to include v4 (but given the track record of this patch,
I totally understand if you want to wait until [2] lands in mips-next and
linux-next).

Rikard

[0]: https://lore.kernel.org/patchwork/patch/1139777/
[1]: https://lore.kernel.org/patchwork/patch/1139778/
[2]: https://lore.kernel.org/linux-mips/20191022192547.480095-1-rikard.falkeborn@xxxxxxxxx/T/#u