[PATCH] [asan] Print SHADOW_SCALE and SHADOW_GRANULARITY as decimal values.

[mats petersson]

Yes, casting to int would be another possible solution - the point is that
the value type should match the format. I don't know enough about this
piece of code which of the two is "more correct". I did spend most of last
year encoding/decoding various printf format strings, so I spotted that
"it's probably wrong".

--
Mats

On 23 April 2015 at 16:12, Daniel Sanders <Daniel.Sanders at imgtec.com> wrote:

>  Good point. Timur has asked me to cast to int and is wondering whether
> the prototype should be changed so I'll wait to see what Kostya says before
> updating the patch to either cast to int or use '%zd'.
>
>
>
> *From:* mats.o.petersson at googlemail.com [mailto:
> mats.o.petersson at googlemail.com] *On Behalf Of *mats petersson
> *Sent:* 23 April 2015 15:41
> *To:* reviews+D9224+public+0db905cdbbbc4cdf at reviews.llvm.org
> *Cc:* Daniel Sanders; kcc at google.com; timurrrr at google.com;
> llvm-commits at cs.uiuc.edu; Sagar Thakur
> *Subject:* Re: [PATCH] [asan] Print SHADOW_SCALE and SHADOW_GRANULARITY
> as decimal values.
>
>
>
>
>
>
>
> On 23 April 2015 at 15:32, Daniel Sanders <daniel.sanders at imgtec.com>
> wrote:
>
> Hi kcc, timurrrr,
>
> During the review of http://reviews.llvm.org/D9199 where I had originally
> changed the debug_mapping.cc test to accept hexadecimal values, we realized
> that SHADOW_SCALE and SHADOW_GRANULARITY ought to be printed as decimal
> values.
> This patch makes that change.
>
> This patch also adds a '0x' prefix to the SHADOW_OFFSET to make it clear
> that
> it is hexadecimal while the other two are decimal.
>
> http://reviews.llvm.org/D9224
>
> Files:
>   lib/asan/asan_rtl.cc
>   test/asan/TestCases/debug_mapping.cc
>
> Index: lib/asan/asan_rtl.cc
> ===================================================================
> --- lib/asan/asan_rtl.cc
> +++ lib/asan/asan_rtl.cc
> @@ -347,9 +347,9 @@
>    Printf("malloc_context_size=%zu\n",
>           (uptr)common_flags()->malloc_context_size);
>
> -  Printf("SHADOW_SCALE: %zx\n", (uptr)SHADOW_SCALE);
> -  Printf("SHADOW_GRANULARITY: %zx\n", (uptr)SHADOW_GRANULARITY);
> -  Printf("SHADOW_OFFSET: %zx\n", (uptr)SHADOW_OFFSET);
> +  Printf("SHADOW_SCALE: %d\n", (uptr)SHADOW_SCALE);
>
> If it's cast to uptr, then surely it should be "%zd" or even "%zu", not
> "%d" (the z means "size_t" length of the value)
>
> --
>
> Mats
>
> +  Printf("SHADOW_GRANULARITY: %d\n", (uptr)SHADOW_GRANULARITY);
> +  Printf("SHADOW_OFFSET: 0x%zx\n", (uptr)SHADOW_OFFSET);
>    CHECK(SHADOW_SCALE >= 3 && SHADOW_SCALE <= 7);
>    if (kMidMemBeg)
>      CHECK(kMidShadowBeg > kLowShadowEnd &&
>
>  Index: test/asan/TestCases/debug_mapping.cc
> ===================================================================
> --- test/asan/TestCases/debug_mapping.cc
> +++ test/asan/TestCases/debug_mapping.cc
> @@ -8,14 +8,14 @@
>
>  // printed because of verbosity=1
>  // CHECK: SHADOW_SCALE: [[SCALE:[0-9]+]]
> -// CHECK: SHADOW_OFFSET: [[OFFSET:[0-9a-f]+]]
> +// CHECK: SHADOW_OFFSET: [[OFFSET:0x[0-9a-f]+]]
>
>  int main() {
>    size_t scale, offset;
>    __asan_get_shadow_mapping(&scale, &offset);
>
> -  fprintf(stderr, "scale: %lx\n", scale);
> -  fprintf(stderr, "offset: %lx\n", offset);
> +  fprintf(stderr, "scale: %d\n", scale);
>
>  Likewise size_t should be "%zd" [and for the hex on the next line "%zx",
> not "%lx"]
>
> +  fprintf(stderr, "offset: 0x%lx\n", offset);
>
>    // CHECK: scale: [[SCALE]]
>    // CHECK: offset: [[OFFSET]]
>
> EMAIL PREFERENCES
>   http://reviews.llvm.org/settings/panel/emailpreferences/
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150423/b321c42c/attachment.html>