It turns out that gcc has real trouble merging all the temporary
on-stack buffer allocation. So despite the fact that their lifetimes do
not overlap, gcc will allocate stack for all of them when they have
different types. Which they do in the number scanning test routines.
This is unfortunate in general, but with lots of test-cases in one
function, it becomes a real problem. gcc will allocate a huge stack
frame for no actual good reason.
We have tried to counteract this tendency of gcc not merging stack slots
(see "-fconserve-stack"), but that has limited effect (and should be on
by default these days, iirc).
So with all the debug options enabled on an i386 allmodconfig build, we
end up with overly big stack frames, and the resulting stack frame size
warnings (now errors):
lib/test_scanf.c: In function ‘numbers_list_field_width_val_width’:
lib/test_scanf.c:530:1: error: the frame size of 2088 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]
530 | }
| ^
lib/test_scanf.c: In function ‘numbers_list_field_width_typemax’:
lib/test_scanf.c:488:1: error: the frame size of 2568 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]
488 | }
| ^
lib/test_scanf.c: In function ‘numbers_list’:
lib/test_scanf.c:437:1: error: the frame size of 2088 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]
437 | }
| ^
In this particular case, the reasonably straightforward solution is to
just split out the test routines into multiple more targeted versions.
That way we don't have one huge stack, but several smaller ones, and
they aren't active all at the same time.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
UBSAN reported (via LKP)
[ 11.021349][ T1] UBSAN: shift-out-of-bounds in lib/test_scanf.c:275:51
[ 11.022782][ T1] shift exponent 32 is too large for 32-bit type 'unsigned int'
When n_bits == 0, the shift is out of range. Switch code to use GENMASK
to handle this case.
Fixes: 50f530e176 ("lib: test_scanf: Add tests for sscanf number conversion")
Reported-by: kernel test robot <oliver.sang@intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Signed-off-by: Petr Mladek <pmladek@suse.com>
Link: https://lore.kernel.org/r/20210727150132.28920-1-andriy.shevchenko@linux.intel.com
sparse was producing warnings of the form:
sparse: cast truncates bits from constant value (ffff0001 becomes 1)
There is no actual problem here. Using type_min() on an unsigned type
results in an (expected) truncation.
However, there is no need to test an unsigned value against type_min().
The minimum value of an unsigned is obviously 0, and any value cast to
an unsigned type is >= 0, so for unsigneds only type_max() need be tested.
This patch also takes the opportunity to clean up the implementation of
simple_numbers_loop() to use a common pattern for the positive and
negative test.
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Signed-off-by: Petr Mladek <pmladek@suse.com>
Link: https://lore.kernel.org/r/20210525122012.6336-2-rf@opensource.cirrus.com
Adds test_sscanf to test various number conversion cases, as
number conversion was previously broken.
This also tests the simple_strtoxxx() functions exported from
vsprintf.c.
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Petr Mladek <pmladek@suse.com>
Link: https://lore.kernel.org/r/20210514161206.30821-3-rf@opensource.cirrus.com
Linus Torvalds
Petr Mladek
Andy Shevchenko
Zhen Lei
Richard Fitzgerald