[PATCH 0/2] sparse v0.6.4 regression

[PATCH 0/2] sparse v0.6.4 regression

[Ramsay Jones]

Hi Luc,

I noticed a regression introduced by commit a69f8d70 which, effectively,
disables the 'memcpy-max-count' check. (I had done a 'make CC=cgcc' in
git, expecting to see some warnings, including the 'memcpy-max-count'
check, and it didn't appear ... :( ). Sorry for not noticing before, but
the 'sparse' make target is supposed to suppress that warning for the
pack-revindex.c file, so ... ;P

[I tried finding your last 'kernel sparse warnings change' list email
to see how many of these warnings still appear in the kernel ... well
it would be zero with v0.6.4!]

I also included the 'case label' patch I sent before, this time with
a commit message and a test. (The test could be extended when sparse
is updated to include the full label-positioning of C23!).

ATB,
Ramsay Jones

Ramsay Jones (2):
  sparse: fix broken 'memcpy-max-count' check
  parse: warn about a 'case label' on empty statement

 parse.c                        |  5 +++++
 sparse.c                       |  2 +-
 validation/byte-count-max.c    | 28 ++++++++++++++++++++++++++++
 validation/label-positioning.c | 22 ++++++++++++++++++++++
 4 files changed, 56 insertions(+), 1 deletion(-)
 create mode 100644 validation/byte-count-max.c
 create mode 100644 validation/label-positioning.c

-- 
2.33.0

Re: [PATCH 0/2] sparse v0.6.4 regression

[Luc Van Oostenryck]

On Tue, Sep 28, 2021 at 12:41:27AM +0100, Ramsay Jones wrote:
> 
> Hi Luc,

Hi Ramsay,

> I noticed a regression introduced by commit a69f8d70 which, effectively,
> disables the 'memcpy-max-count' check. (I had done a 'make CC=cgcc' in
> git, expecting to see some warnings, including the 'memcpy-max-count'
> check, and it didn't appear ... :( ). Sorry for not noticing before, but
> the 'sparse' make target is supposed to suppress that warning for the
> pack-revindex.c file, so ... ;P
> 
> [I tried finding your last 'kernel sparse warnings change' list email
> to see how many of these warnings still appear in the kernel ... well
> it would be zero with v0.6.4!]

Now I can see 4 of them.
 
> I also included the 'case label' patch I sent before, this time with
> a commit message and a test. (The test could be extended when sparse
> is updated to include the full label-positioning of C23!).

Thank you very much, pushed now, and ...
apologies for this huge delay.

-- Luc 

Re: [PATCH 0/2] sparse v0.6.4 regression

[Ramsay Jones]

On 21/05/2022 10:44, Luc Van Oostenryck wrote:
> On Tue, Sep 28, 2021 at 12:41:27AM +0100, Ramsay Jones wrote:

> Hi Ramsay,

Hi Luc, good to hear from you! Hope you are keeping well.

>> I noticed a regression introduced by commit a69f8d70 which, effectively,
>> disables the 'memcpy-max-count' check. (I had done a 'make CC=cgcc' in
>> git, expecting to see some warnings, including the 'memcpy-max-count'
>> check, and it didn't appear ... :( ). Sorry for not noticing before, but
>> the 'sparse' make target is supposed to suppress that warning for the
>> pack-revindex.c file, so ... ;P
>>
>> [I tried finding your last 'kernel sparse warnings change' list email
>> to see how many of these warnings still appear in the kernel ... well
>> it would be zero with v0.6.4!]
> 
> Now I can see 4 of them.

I tested today's master (@commit fbdc046e (Use offsetof macro to silence
null ptr subtraction warning, 2022-03-21)), without issue on 64-bit
linux and cygwin. (so, not 32-bit linux, which requires a reboot ...)

Also, a 'make CC=cgcc' now issues all the warnings I expect, including:

      CC pack-revindex.o
  pack-revindex.c:73:23: warning: memset with byte count of 262144

I noticed that three RISC-V patches have been pushed since then, so I will
test those later ...

>> I also included the 'case label' patch I sent before, this time with
>> a commit message and a test. (The test could be extended when sparse
>> is updated to include the full label-positioning of C23!).
> 
> Thank you very much, pushed now, and ...

Thanks!

> apologies for this huge delay.

No problem.

ATB,
Ramsay Jones