Skip to contentSkip to conversation

Fix some "enum value 'SYM_...' not handled in switch" warnings

PATCH Applied Ramsay Jones 3 participants 7 messages Jul 18, 2009 – Jul 24, 2009 Sparse checker

lore
Share & tools
lore
#0 lore raw

PATCH 2/5 Fix some "enum value 'SYM_...' not handled in switch" warnings

---

Hi Chris,

These warnings were issued by gcc v3.4.4, but not by gcc v4.1.2. So I guess gcc probably found these warnings to be too noisy ...

ATB,
Ramsay Jones

 parse.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
parse.c +2 -0
index e5ad867..42e8c74 100644
@@ -2616,6 +2616,8 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
 			case SYM_ENUM:
 			case SYM_RESTRICT:
 				base_type->ident = ident;
+			default:
+				break;
 			}
 		}
 	} else if (base_type && base_type->type == SYM_FN) {
Signature · 1 line
1.6.3.1
#1 lore raw
Quoted 2 lines · Ramsay Jones

These warnings were issued by gcc v3.4.4, but not by gcc v4.1.2. So I guess gcc probably found these warnings to be too noisy ...

[...]

parse.c struct token *external_declaration(struct token *token, struct symbol_list **lis · line 2621

Quoted 9 lines
parse.c +2 -0
@@ -2616,6 +2616,8 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
 			case SYM_ENUM:
 			case SYM_RESTRICT:
 				base_type->ident = ident;
+			default:
+				break;
 			}

I don't think you want to add defaults like this just to avoid warnings. Warnings like that can help when adding a new item to an enum, to find the places where you need to extend the code to hand the new item. And since current GCC doesn't even issue the warning by default, it seems even more unnecessary to add that default case.

- Josh Triplett

#2 lore raw
Quoted 4 lines · Josh Triplett
Quoted 2 lines · Ramsay Jones

These warnings were issued by gcc v3.4.4, but not by gcc v4.1.2. So I guess gcc probably found these warnings to be too noisy ...

[...]

Quoted 16 lines

parse.c struct token *external_declaration(struct token *token, struct symbol_list **lis · line 2621

Quoted 9 lines
parse.c +2 -0
@@ -2616,6 +2616,8 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
 			case SYM_ENUM:
 			case SYM_RESTRICT:
 				base_type->ident = ident;
+			default:
+				break;
 			}

I don't think you want to add defaults like this just to avoid warnings. Warnings like that can help when adding a new item to an enum, to find the places where you need to extend the code to hand the new item. And since current GCC doesn't even issue the warning by default, it seems even more unnecessary to add that default case.

OK...

So, if I understand your argument, in order to make the best use of these warnings, then the correct change would look like the diff given below, and (for more up-to-date gcc) add -Wswitch-enum to CFLAGS (at least occasionally).

Yes?

[if no, then I obviously don't understand ;-)]

Hmm, I have another idea; a new patch is brewing!

ATB,
Ramsay Jones

-->8--

parse.c +17 -0
index e5ad867..afe39c3 100644
@@ -2616,6 +2616,23 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
 			case SYM_ENUM:
 			case SYM_RESTRICT:
 				base_type->ident = ident;
+				break;
+			case SYM_UNINITIALIZED:
+			case SYM_PREPROCESSOR:
+			case SYM_BASETYPE:
+			case SYM_NODE:
+			case SYM_PTR:
+			case SYM_FN:
+			case SYM_ARRAY:
+			case SYM_TYPEDEF:
+			case SYM_TYPEOF:
+			case SYM_MEMBER:
+			case SYM_BITFIELD:
+			case SYM_LABEL:
+			case SYM_FOULED:
+			case SYM_KEYWORD:
+			case SYM_BAD:
+				break;
 			}
 		}
 	} else if (base_type && base_type->type == SYM_FN) {

-->8--

maintainer
#3 lore raw
Quoted 7 lines · Ramsay Jones
Quoted 6 lines

I don't think you want to add defaults like this just to avoid warnings. Warnings like that can help when adding a new item to an enum, to find the places where you need to extend the code to hand the new item.  And since current GCC doesn't even issue the warning by default, it seems even more unnecessary to add that default case.

Quoted 6 lines

OK...

So, if I understand your argument, in order to make the best use of these warnings, then the correct change would look like the diff given below, and (for more up-to-date gcc) add -Wswitch-enum to CFLAGS (at least occasionally).

OK. I don't want to list all the enumerate value here just for the sake of gcc warnings.
It makes the code ugly. Nor do I want to change the gcc flags used to compile sparse. If the newest gcc still complain on those. I would rather add the blank default
to make it clean. Since latest gcc doesn't issue warning on those. I think it is fine to
leave it as it is.

Chris
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#4 lore raw
Quoted 9 lines · Christopher Li

OK. I don't want to list all the enumerate value here just for the sake of gcc warnings.
It makes the code ugly. Nor do I want to change the gcc flags used to compile sparse. If the newest gcc still complain on those. I would rather add the blank default
to make it clean. Since latest gcc doesn't issue warning on those. I think it is fine to
leave it as it is.

OK, no problem. I can maintain this in my cygwin repo, along with the __sentinel__ attribute patch. (As before, I don't need this on Linux).

Just FYI, I had planned to fix this by implementing a local makefile config file (like config.mak in git), so that I could fix this issue locally ("out of tree").

However, I was (pleasantly) surprised to find that I'd been beaten to the punch by Samuel Bronson in commit 8d86d0e.

So I tried this local.mk file:

    OS=cygwin
    CFLAGS+=-Wno-switch-enum

This lead to the invocation of gcc I was hoping for; something like:

    gcc ... -Wall ... -Wno-switch-enum parse.c

which I had expected to suppress the warnings. It didn't. *Ho-Hum* ;-)

ATB,
Ramsay Jones

maintainer
#5 lore raw
Quoted 7 lines · Ramsay Jones

Just FYI, I had planned to fix this by implementing a local makefile config file (like config.mak in git), so that I could fix this issue locally ("out of tree").

However, I was (pleasantly) surprised to find that I'd been beaten to the punch by Samuel Bronson in commit 8d86d0e.

Quoted 8 lines

So I tried this local.mk file:

   OS=cygwin
   CFLAGS+=-Wno-switch-enum

This lead to the invocation of gcc I was hoping for; something like:

   gcc ... -Wall ... -Wno-switch-enum parse.c

Yes, that works. I try the latest version of the cygwin. The gcc in cygwin is still old enough to give the warning. I am going to leave it as it is.

BTW, you can put some local build target in local.mk as well, not just config changes.

Chris
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#6 lore raw
In reply to Ramsay Jones
Quoted 28 lines · Ramsay Jones
Quoted 4 lines · Josh Triplett
Quoted 2 lines · Ramsay Jones

These warnings were issued by gcc v3.4.4, but not by gcc v4.1.2. So I guess gcc probably found these warnings to be too noisy ...

[...]

Quoted 16 lines

parse.c struct token *external_declaration(struct token *token, struct symbol_list **lis · line 2621

Quoted 9 lines
parse.c +2 -0
@@ -2616,6 +2616,8 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
 			case SYM_ENUM:
 			case SYM_RESTRICT:
 				base_type->ident = ident;
+			default:
+				break;
 			}

I don't think you want to add defaults like this just to avoid warnings. Warnings like that can help when adding a new item to an enum, to find the places where you need to extend the code to hand the new item. And since current GCC doesn't even issue the warning by default, it seems even more unnecessary to add that default case.

OK...

So, if I understand your argument, in order to make the best use of these warnings, then the correct change would look like the diff given below, and (for more up-to-date gcc) add -Wswitch-enum to CFLAGS (at least occasionally).

[...]

Quoted 15 lines
parse.c +7 -0
index e5ad867..afe39c3 100644
@@ -2616,6 +2616,23 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
 			case SYM_ENUM:
 			case SYM_RESTRICT:
 				base_type->ident = ident;
+				break;
+			case SYM_UNINITIALIZED:
+			case SYM_PREPROCESSOR:
+			case SYM_BASETYPE:
+			case SYM_NODE:
+			case SYM_PTR:
+			case SYM_FN:

parse.c struct token *external_declaration(struct token *token, struct symbol_list **lis · line 2635

Quoted 10 lines
parse.c +10 -0
+			case SYM_ARRAY:
+			case SYM_TYPEDEF:
+			case SYM_TYPEOF:
+			case SYM_MEMBER:
+			case SYM_BITFIELD:
+			case SYM_LABEL:
+			case SYM_FOULED:
+			case SYM_KEYWORD:
+			case SYM_BAD:
+				break;

This represents exactly why more recent GCC defaults this warning to off. ;)

No, I don't think writing the code this way helps. I think it makes more sense to leave the warning off and not bother placating older versions of GCC.

- Josh Triplett