On Thu, 12 Nov 2015 16:26:27 +0100 Stefan Schmidt <ste...@osg.samsung.com> said:
> Hello. > > I took a moment to run our code through the sparse semantic parser. > https://sparse.wiki.kernel.org/index.php/Main_Page > > You can easily do so as well by installing sparse and run "make CC=cgcc" > to enable the build wrapper. > > But now to some actual findings. Really most of it has been noise and i > reduced it to some things here manually. > > Find some remarks on some of the categories inside. If you feel like > this is on your turf and you either think its false positive or you are > going to fix it let me know. > > lib/eina/eina_fp.c:14:4: warning: constant 0x0000000100000000 is so big > it is long > lib/eina/eina_fp.c:439:17: warning: constant 0x00000003243f6a89 is so > big it is long > lib/eina/eina_fp.c:440:17: warning: constant 0x00000003243f6a89 is so > big it is long > lib/eina/eina_fp.c:441:18: warning: constant 0x00000003243f6a89 is so > big it is long > lib/eina/eina_fp.c:447:28: warning: constant 0x00000003243f6a89 is so > big it is long > lib/eina/eina_fp.c:447:46: warning: constant 0x00000003243f6a89 is so > big it is long > lib/eina/eina_fp.c:453:31: warning: constant 0x00000003243f6a89 is so > big it is long > lib/eina/eina_fp.c:479:17: warning: constant 0x00000003243f6a89 is so > big it is long > lib/ecore/ecore_anim.c:583:16: warning: constant 0x00000003243f6a89 is > so big it is long > lib/ecore/ecore_anim.c:583:16: warning: constant 0x00000003243f6a89 is > so big it is long > lib/ecore/ecore_anim.c:588:16: warning: constant 0x00000003243f6a89 is > so big it is long > lib/ecore/ecore_anim.c:593:16: warning: constant 0x00000003243f6a89 is > so big it is long > > ----- > > lib/eina/eina_rbtree.c:63:33: error: dubious one-bit signed bitfield > lib/eina/eina_thread_queue.c:62:42: error: dubious one-bit signed bitfield > lib/evas/canvas/evas_object_text.c:71:43: error: dubious one-bit signed > bitfield > lib/evas/canvas/evas_object_text.c:72:46: error: dubious one-bit signed > bitfield > modules/evas/engines/buffer/evas_engine.h:47:51: error: dubious one-bit > signed bitfield > modules/evas/engines/buffer/evas_engine.h:48:49: error: dubious one-bit > signed bitfield > modules/evas/engines/buffer/Evas_Engine_Buffer.h:24:30: error: dubious > one-bit signed bitfield > modules/evas/engines/buffer/evas_engine.h:47:51: error: dubious one-bit > signed bitfield > modules/evas/engines/buffer/evas_engine.h:48:49: error: dubious one-bit > signed bitfield > ../src/modules/evas/engines/buffer/Evas_Engine_Buffer.h:24:30: error: > dubious one-bit signed bitfield > ../src/modules/evas/engines/buffer/Evas_Engine_Buffer.h:24:30: error: > dubious one-bit signed bitfield > ../src/modules/evas/engines/buffer/Evas_Engine_Buffer.h:24:30: error: > dubious one-bit signed bitfield > ../src/modules/evas/engines/buffer/Evas_Engine_Buffer.h:24:30: error: > dubious one-bit signed bitfield > lib/ecore_ipc/ecore_ipc_private.h:84:35: error: dubious one-bit signed > bitfield > lib/ecore_ipc/ecore_ipc_private.h:102:35: error: dubious one-bit signed > bitfield > modules/ecore_evas/engines/x/ecore_evas_x.c:902:29: error: dubious > one-bit signed bitfield > modules/ecore_evas/engines/x/ecore_evas_x.c:903:33: error: dubious > one-bit signed bitfield > modules/ecore_evas/engines/x/ecore_evas_x.c:904:30: error: dubious > one-bit signed bitfield > modules/ecore_evas/engines/x/ecore_evas_x.c:905:34: error: dubious > one-bit signed bitfield > modules/ecore_evas/engines/x/ecore_evas_x.c:906:34: error: dubious > one-bit signed bitfield > modules/evas/image_loaders/gif/evas_image_load_gif.c:37:23: error: > dubious one-bit signed bitfield > > Funny one. This bitfields are signed so they can only hold -1 and 0. Not > a real bummer as we normally use them in if (x) only but if we have > cases where we really expect an 1 after we might be in trouble. Could be > a good idea wot move them over to Eina_Bool. not sure it makes any difference. if it's a 1 bit field... it will always be used with if (x) or if (!x) ... setting it to 1 yes - will set it to -1, but we don't care really. :) well ok - the way i use bitfields like this it doesn't matter, as i would never do if (x == 1) or if (x == 0) on them - ever. it's far too long for me :) i get why it MAY be a problem (like this example) but it's stupid code to write imho... > ----- > > lib/emile/emile_main.c:149:53: warning: right shift by bigger than > source value > lib/emile/emile_main.c:150:53: warning: right shift by bigger than > source value > lib/emile/emile_main.c:151:53: warning: right shift by bigger than > source value > > Cedric? > > ----- > > static_libs/rg_etc/rg_etc1.c:1444:21: warning: crazy programmer > static_libs/rg_etc/rg_etc1.c:1458:21: warning: crazy programmer > static_libs/rg_etc/rg_etc1.c:1483:21: warning: crazy programmer > modules/evas/image_savers/tgv/evas_image_save_tgv.c:49:14: warning: > crazy programmer > > Man, I love this warnings. :) i detect a cedric! :) > ----- > > lib/ecore_input/ecore_input_compose.h:9900:19: warning: trying to > concatenate 63013-character string (8191 bytes max) i'm rather proud of this bit of pure evil in efl. :) > modules/evas/engines/gl_common/shader_3d/evas_gl_3d_shaders.x:975:9: > warning: trying to concatenate 8394-character string (8191 bytes max) > modules/evas/engines/gl_common/shader_3d/evas_gl_3d_shaders.x:1441:9: > warning: trying to concatenate 10118-character string (8191 bytes max) > modules/evas/engines/gl_common/shader_3d/evas_gl_3d_shaders.x:2113:9: > warning: trying to concatenate 11086-character string (8191 bytes max) > > What are we going to do about this? Do we take our maximal size for > strings seriously or does this not really matter? as i read it compilers MUST provide at LEAST 2048 bytes for such strings (c89) and i think c99 raised it as a minimum to 4096. i have no idea where this 8192 limit comes from (8191 chars plus 0 byte). as such we have had this almost 65k string in efl for years now and it has never hit a problem - it should at compile time (compiler should overflow whatever buffer size it has and barf)... unless the compiler is majorly sucky. maybe 8192 was just a number thrown into this tool as a "really long string" number to be worried about if the string is that big? ie - maybe you didn't mean this to happen? > ----- > > lib/evas/canvas/evas_map.c:284:27: warning: Variable length array is used. this one is indeed dubious. it's declaring a variable size array based on local vars at the time of the call. this should become an alloca. likely for all of these too (haven't looked). > lib/ector/software/ector_software_rasterizer.c:215:22: warning: Variable > length array is used. > lib/ector/software/ector_software_rasterizer.c:250:22: warning: Variable > length array is used. > lib/ecore_x/xlib/ecore_x_randr.c:559:42: warning: Variable length array > is used. > modules/evas/engines/gl_common/evas_gl_3d_shader.c:181:20: warning: > Variable length array is used. > bin/evas/evas_cserve2_shm_debug.c:42:14: warning: Variable length array > is used. > > ----- > > lib/emotion/emotion_webcam.c:120:27: warning: dubious: !x & y > > ----- > > modules/evas/engines/gl_common/evas_gl_core.c:1624:46: error: too many > arguments for function glsym_evas_gl_native_context_get > modules/evas/engines/gl_common/evas_gl_core.c:1638:40: error: too many > arguments for function glsym_evas_gl_engine_data_get > > Seems to be a misunderstanding of the function pointer. yeah. we have an unusual () as the args - in c this means (0 OR more arguments). you can actually pass anything you like... you don't get any param/type checking then though. > ----- > > modules/evas/engines/gl_x11/evas_engine.c:230:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:285:44: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:310:44: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:329:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:358:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:374:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:381:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:404:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:446:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:474:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:481:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:557:81: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:558:66: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:576:48: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:577:53: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:599:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:632:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:641:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:850:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:857:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:895:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:902:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:923:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:1127:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:1145:39: error: too many > arguments for function glsym_evas_gl_common_error_set > modules/evas/engines/gl_x11/evas_engine.c:1545:40: error: too many > arguments for function glsym_evas_gl_common_shaders_flush > modules/evas/engines/gl_x11/evas_engine.c:1967:45: error: too many > arguments for function glsym_evas_gl_common_error_get > modules/evas/engines/gl_x11/evas_engine.c:1982:34: error: too many > arguments for function glsym_evas_gl_common_error_set same for this as above. making use of () == 0 or more params. the tool doesn't understand this bit of c so it is tehcnically wrong. :) > ----- > > lib/ecore_wayland/ecore_wl_input.c:128:4: warning: incorrect type in > initializer (incompatible argument 3 (different signedness)) > lib/ecore_wayland/ecore_wl_input.c:128:4: expected void ( > *capabilities )( ... ) > lib/ecore_wayland/ecore_wl_input.c:128:4: got void ( static > [toplevel] *<noident> )( ... ) > > Have fun with these. > > regards > Stefan Schmidt > > ------------------------------------------------------------------------------ > _______________________________________________ > enlightenment-devel mailing list > enlightenment-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/enlightenment-devel > -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) ras...@rasterman.com ------------------------------------------------------------------------------ _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
