stadtx_hash.h: Silence -Wimplicit-fallthrough compilation warnings #16206
Comments
From @jkeenanWhen building perl with gcc-7.2, many new compiler warnings are thrown. http://perl5.test-smoke.org/report/58805 Those warnings come from these files: ##### The patch attached silences the 'warning: this statement may fall The following warning remains during a gcc-7.2 build: ##### I don't yet know how to address that one. Please review the patch. Thank you very much. |
From @jkeenan0001-Prevent-warnings-when-building-with-gcc-7.2.patchFrom 915d1758bac3d1a7e1b7ec6c570ef70fce99d279 Mon Sep 17 00:00:00 2001
From: James E Keenan <[email protected]>
Date: Sat, 21 Oct 2017 21:27:18 -0400
Subject: [PATCH] Prevent warnings when building with gcc-7.2.
Guard against: "warning: this statement may fall through" warnings.
---
stadtx_hash.h | 22 ++++++++++++++++++++++
1 file changed, 22 insertions(+)
diff --git a/stadtx_hash.h b/stadtx_hash.h
index a54af2e..3b5dfb6 100644
--- a/stadtx_hash.h
+++ b/stadtx_hash.h
@@ -198,29 +198,39 @@ STADTX_STATIC_INLINE U64 stadtx_hash_with_state(
v0= ROTR64(v0, 17) ^ v1;
v1= ROTR64(v1, 53) + v0;
key += 8;
+ /* FALLTHROUGH */
case 2:
v0 += U8TO64_LE(key) * STADTX_K3_U64;
v0= ROTR64(v0, 17) ^ v1;
v1= ROTR64(v1, 53) + v0;
key += 8;
+ /* FALLTHROUGH */
case 1:
v0 += U8TO64_LE(key) * STADTX_K3_U64;
v0= ROTR64(v0, 17) ^ v1;
v1= ROTR64(v1, 53) + v0;
key += 8;
+ /* FALLTHROUGH */
case 0:
default: break;
}
switch ( len & 0x7 ) {
case 7: v0 += (U64)key[6] << 32;
+ /* FALLTHROUGH */
case 6: v1 += (U64)key[5] << 48;
+ /* FALLTHROUGH */
case 5: v0 += (U64)key[4] << 16;
+ /* FALLTHROUGH */
case 4: v1 += (U64)U8TO32_LE(key);
break;
+ /* FALLTHROUGH */
case 3: v0 += (U64)key[2] << 48;
+ /* FALLTHROUGH */
case 2: v1 += (U64)U8TO16_LE(key);
break;
+ /* FALLTHROUGH */
case 1: v0 += (U64)key[0];
+ /* FALLTHROUGH */
case 0: v1 = ROTL64(v1, 32) ^ 0xFF;
break;
}
@@ -253,25 +263,37 @@ STADTX_STATIC_INLINE U64 stadtx_hash_with_state(
switch ( len >> 3 ) {
case 3: v0 += ((U64)U8TO64_LE(key) * STADTX_K2_U32); key += 8; v0= ROTL64(v0,57) ^ v3;
+ /* FALLTHROUGH */
case 2: v1 += ((U64)U8TO64_LE(key) * STADTX_K3_U32); key += 8; v1= ROTL64(v1,63) ^ v2;
+ /* FALLTHROUGH */
case 1: v2 += ((U64)U8TO64_LE(key) * STADTX_K4_U32); key += 8; v2= ROTR64(v2,47) + v0;
+ /* FALLTHROUGH */
case 0: v3 = ROTR64(v3,11) - v1;
+ /* FALLTHROUGH */
}
v0 ^= (len+1) * STADTX_K3_U64;
switch ( len & 0x7 ) {
case 7: v1 += (U64)key[6];
+ /* FALLTHROUGH */
case 6: v2 += (U64)U8TO16_LE(key+4);
v3 += (U64)U8TO32_LE(key);
break;
+ /* FALLTHROUGH */
case 5: v1 += (U64)key[4];
+ /* FALLTHROUGH */
case 4: v2 += (U64)U8TO32_LE(key);
break;
+ /* FALLTHROUGH */
case 3: v3 += (U64)key[2];
+ /* FALLTHROUGH */
case 2: v1 += (U64)U8TO16_LE(key);
break;
+ /* FALLTHROUGH */
case 1: v2 += (U64)key[0];
+ /* FALLTHROUGH */
case 0: v3 = ROTL64(v3, 32) ^ 0xFF;
break;
+ /* FALLTHROUGH */
}
v1 -= v2;
--
2.7.4
|
From @jkeenanSummary of my perl5 (revision 5 version 27 subversion 6) configuration: Characteristics of this binary (from libperl): |
From @jkeenanOn Sun, 22 Oct 2017 02:25:29 GMT, jkeenan@pobox.com wrote:
Code review by mauke++ on #p5p indicated that some of my FALLTHROUGH guards were superfluous. Providing supplementary patch. Please review. Thank you very much. |
From @jkeenan0002-Remove-superfluous-FALLTHROUGH-guards.patchFrom 4fd1c57d681355f0e0bb51ead65c9552fbaf325a Mon Sep 17 00:00:00 2001
From: James E Keenan <[email protected]>
Date: Sun, 22 Oct 2017 07:48:43 -0400
Subject: [PATCH 2/2] Remove superfluous FALLTHROUGH guards.
Per code review by mauke++ on #p5p.
---
stadtx_hash.h | 6 ------
1 file changed, 6 deletions(-)
diff --git a/stadtx_hash.h b/stadtx_hash.h
index 3b5dfb6..c755018 100644
--- a/stadtx_hash.h
+++ b/stadtx_hash.h
@@ -223,12 +223,10 @@ STADTX_STATIC_INLINE U64 stadtx_hash_with_state(
/* FALLTHROUGH */
case 4: v1 += (U64)U8TO32_LE(key);
break;
- /* FALLTHROUGH */
case 3: v0 += (U64)key[2] << 48;
/* FALLTHROUGH */
case 2: v1 += (U64)U8TO16_LE(key);
break;
- /* FALLTHROUGH */
case 1: v0 += (U64)key[0];
/* FALLTHROUGH */
case 0: v1 = ROTL64(v1, 32) ^ 0xFF;
@@ -278,22 +276,18 @@ STADTX_STATIC_INLINE U64 stadtx_hash_with_state(
case 6: v2 += (U64)U8TO16_LE(key+4);
v3 += (U64)U8TO32_LE(key);
break;
- /* FALLTHROUGH */
case 5: v1 += (U64)key[4];
/* FALLTHROUGH */
case 4: v2 += (U64)U8TO32_LE(key);
break;
- /* FALLTHROUGH */
case 3: v3 += (U64)key[2];
/* FALLTHROUGH */
case 2: v1 += (U64)U8TO16_LE(key);
break;
- /* FALLTHROUGH */
case 1: v2 += (U64)key[0];
/* FALLTHROUGH */
case 0: v3 = ROTL64(v3, 32) ^ 0xFF;
break;
- /* FALLTHROUGH */
}
v1 -= v2;
--
2.7.4
|
|
The RT System itself - Status changed from 'new' to 'open' |
From @jkeenanOn Sun, 22 Oct 2017 12:08:01 GMT, jkeenan wrote:
Patches were applied to blead in commit b2733f8 and 6437ba6. Most of the implicit-fallthrough warnings are gone. However, when I subsequently built with gcc-7 on a *threaded build*, there were still some of those warnings: ##### Needs further investigation. Thank you very much. |
From @maukeOn Sun, 22 Oct 2017 06:32:38 -0700, jkeenan wrote:
I believe the fallthrough warning is fixed in commit 146930d. |
From @jkeenanOn Sun, 22 Oct 2017 14:21:58 GMT, mauke- wrote:
Yes -- but unfortunately I just spotted this: ##### -- |
From @jkeenanOn Sun, 22 Oct 2017 14:57:00 GMT, jkeenan wrote:
And mauke++ has taken care of this one as well: ##### get rid of "implicit fallthrough" warnings with gcc 7 I think we've now eliminated this warning from non-threaded and threaded builds on Linux with gcc-7 -- at least from those files that are p5p-maintained. I will prepare patches for files that are cpan-upstream, but we don't need to keep this ticket open for that. Marking ticket Resolved. Thank you very much. |
|
@jkeenan - Status changed from 'open' to 'resolved' |
Migrated from rt.perl.org#132342 (status was 'resolved')
Searchable as RT132342$
The text was updated successfully, but these errors were encountered: