Re: [PATCH v2] lib: zstd: Fix -Wstringop-overflow warning

[Kees Cook <keescook@chromium.org>]

On Sat, Jan 14, 2023 at 01:06:07AM +0000, Nick Terrell wrote:
> 
> 
> > On Jan 13, 2023, at 4:48 PM, Kees Cook <keescook@chromium.org> wrote:
> > 
> > !-------------------------------------------------------------------|
> >  This Message Is From an External Sender
> > 
> > |-------------------------------------------------------------------!
> > 
> > On Tue, Jan 10, 2023 at 11:10:08PM +0000, Nick Terrell wrote:
> >> 
> >> 
> >>> On Jan 4, 2023, at 1:20 PM, Kees Cook <keescook@chromium.org> wrote:
> >>> 
> >>> !-------------------------------------------------------------------|
> >>> This Message Is From an External Sender
> >>> 
> >>> |-------------------------------------------------------------------!
> >>> 
> >>> Fix the following -Wstringop-overflow warning when building with GCC 11+:
> >>> 
> >>> lib/zstd/decompress/huf_decompress.c: In function ‘HUF_readDTableX2_wksp’:
> >>> lib/zstd/decompress/huf_decompress.c:700:5: warning: ‘HUF_fillDTableX2.constprop’ accessing 624 bytes in a region of size 52 [-Wstringop-overflow=]
> >>> 700 |     HUF_fillDTableX2(dt, maxTableLog,
> >>>     |     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >>> 701 |                    wksp->sortedSymbol, sizeOfSort,
> >>>     |                    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >>> 702 |                    wksp->rankStart0, wksp->rankVal, maxW,
> >>>     |                    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >>> 703 |                    tableLog+1,
> >>>     |                    ~~~~~~~~~~~
> >>> 704 |                    wksp->calleeWksp, sizeof(wksp->calleeWksp) / sizeof(U32));
> >>>     |
> >>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >>> lib/zstd/decompress/huf_decompress.c:700:5: note: referencing argument 6 of type ‘U32 (*)[13]’ {aka ‘unsigned int (*)[13]’}
> >>> lib/zstd/decompress/huf_decompress.c:571:13: note: in a call to function ‘HUF_fillDTableX2.constprop’
> >>> 571 | static void HUF_fillDTableX2(HUF_DEltX2* DTable, const U32 targetLog,
> >>>     |             ^~~~~~~~~~~~~~~~
> >>> 
> >>> by using pointer notation instead of array notation.
> >>> 
> >>> This is one of the last remaining warnings to be fixed before globally
> >>> enabling -Wstringop-overflow.
> >> 
> >> The patch looks correct to me, thanks for reviving it. But, I was attempting to reproduce the issue,
> >> so I could better understand what's going on, and I wasn't able to reproduce it myself.
> >> 
> >> To attempt to reproduce, I applied this patch
> >> 
> >> ---
> >> diff --git a/lib/zstd/Makefile b/lib/zstd/Makefile
> >> index 20f08c644b71..190d3d5ab4be 100644
> >> --- a/lib/zstd/Makefile
> >> +++ b/lib/zstd/Makefile
> >> @@ -12,6 +12,8 @@ obj-$(CONFIG_ZSTD_COMPRESS) += zstd_compress.o
> >> obj-$(CONFIG_ZSTD_DECOMPRESS) += zstd_decompress.o
> >> obj-$(CONFIG_ZSTD_COMMON) += zstd_common.o
> >> 
> >> +ccflags-y := -Wstringop-overflow=4 -Werror
> >> +
> >> zstd_compress-y := \
> >>                zstd_compress_module.o \
> >>                compress/fse_compress.o \
> >> ---
> >> 
> >> Then compiled on x86-64 with gcc 12.2.0 on tag v6.2-rc3. I saw no errors.
> >> I also tried with just `-Wstringop-overflow`, and on upstream zstd. I tried to
> >> make a minimal reproducer on godbolt, so I could see if it was the gcc version,
> >> but wasn't able to make it fail with any of them https://gcc.godbolt.org/z/Exzq9arMr .
> >> 
> >> Could you please tell me how to reproduce this warning?
> > 
> > I saw it like so with next-20230113 on x86_64:
> > 
> > $ gcc --version
> > gcc (Ubuntu 12.2.0-3ubuntu1) 12.2.0
> > ...
> > $ make KCFLAGS=-Wstringop-overflow allmodconfig lib/zstd/decompress/huf_decompress.o
> 
> Thanks, I was able to repro it! I will merge this patch into my tree.

Thanks!

> If you would like to submit the same patch upstream yourself, I will accept the PR, otherwise I can submit an upstream PR.

I don't know the process there, so if you could do it, I'd appreciate
it.

> Just to be certain, this patch is to work around a shortcoming in
> -Wstringop-overflow, but the code was otherwise correct?

It's the same result for the binary output. The types indicated in the
function prototype meant there was cross-struct-member overflow (i.e.
writing to the second array dimension when only 1 was specified), but
the way to clear up intention isn't as clean here, so I'd kind of say
half a code correctness issue, and half a work-around. *makes wavey
hands gesture*

-Kees

-- 
Kees Cook