Re: [Cocci] Checking software behaviour according to selected SmPL code variants
>> @display@ >> expression e; >> @@ >> *brelse(e); >> <+... when != e = ... >> (e = ... >> | >> *e >> )...+> >> >> >> I would find it nicer if I do not need to repeat a code exclusion >> specification >> as the first element of a SmPL disjunction for this special use case. > > You don't. The when you have is useless. I am curious under which circumstances we will achieve a better common understanding for the shown application of SmPL disjunctions eventually together with additional code exclusion specifications. Regards, Markus ___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
On Wed, 10 Jun 2020, Markus Elfring wrote: > >> I hoped that my specification of a SmPL code exclusion should prevent > >> the presentation of assignments (independent from statements and/or > >> expressions). > > > > ( > > e = e1 > > | > > *e > > ) > > @display@ > expression e; > @@ > *brelse(e); > <+... when != e = ... > (e = ... > | > *e > )...+> > > > I would find it nicer if I do not need to repeat a code exclusion > specification > as the first element of a SmPL disjunction for this special use case. You don't. The when you have is useless. julia ___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
>> I hoped that my specification of a SmPL code exclusion should prevent >> the presentation of assignments (independent from statements and/or >> expressions). > > ( > e = e1 > | > *e > ) @display@ expression e; @@ *brelse(e); <+... when != e = ... (e = ... | *e )...+> I would find it nicer if I do not need to repeat a code exclusion specification as the first element of a SmPL disjunction for this special use case. Regards, Markus ___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
On Wed, 10 Jun 2020, Markus Elfring wrote: > >> @display@ > >> expression e; > >> @@ > >> *brelse(e); > >> <+... when != e = ... > >> *e > >> ...+> > … > >> elfring@Sonne:~/Projekte/Linux/next-patched> spatch > >> ~/Projekte/Coccinelle/janitor/show_questionable_brelse_usage8.cocci > >> fs/ext4/extents.c > >> … > >> @@ -1127,8 +1121,6 @@ static int ext4_ext_split(handle_t *hand > … > >> - brelse(bh); > >> - bh = NULL; > … > >> Would you like to suggest any fine-tuning for the search approach? > > > > When describes what happens elsewhere than in the statement matched by the > > pattern. > > I hoped that my specification of a SmPL code exclusion should prevent > the presentation of assignments (independent from statements and/or > expressions). ( e = e1 | *e )___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
>> @display@ >> expression e; >> @@ >> *brelse(e); >> <+... when != e = ... >> *e >> ...+> … >> elfring@Sonne:~/Projekte/Linux/next-patched> spatch >> ~/Projekte/Coccinelle/janitor/show_questionable_brelse_usage8.cocci >> fs/ext4/extents.c >> … >> @@ -1127,8 +1121,6 @@ static int ext4_ext_split(handle_t *hand … >> -brelse(bh); >> -bh = NULL; … >> Would you like to suggest any fine-tuning for the search approach? > > When describes what happens elsewhere than in the statement matched by the > pattern. I hoped that my specification of a SmPL code exclusion should prevent the presentation of assignments (independent from statements and/or expressions). Regards, Markus ___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
> That is quite normal. One statement should be followed by another statement. I have tried another script variant out for the semantic patch language. @display@ expression e; @@ *brelse(e); <+... when != e = ... *e ...+> I wonder about the generation of a diff hunk then like the following. elfring@Sonne:~/Projekte/Linux/next-patched> spatch ~/Projekte/Coccinelle/janitor/show_questionable_brelse_usage8.cocci fs/ext4/extents.c … @@ -1127,8 +1121,6 @@ static int ext4_ext_split(handle_t *hand err = ext4_handle_dirty_metadata(handle, inode, bh); if (err) goto cleanup; - brelse(bh); - bh = NULL; /* correct old leaf */ if (m) { … Would you like to suggest any fine-tuning for the search approach? Regards, Markus ___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
>> If I omit the specification “, ...” from the function call parameters >> because I could be unsure about the number of arguments in other >> software situations, I do not get the desired test output as before. > > This has been discussed before. I was just looking for a related update in this area. > When you put <+... ...+> in an argument list, it doesn't mean an unknown > number > of arguments, it means one arguemnt that has something as a subexpression. Another SmPL code variant can eventually become interesting then. *y = x(..., <+... e ...+>, ...) >> If I omit even the semicolon from the assignment statement in the >> search pattern, I get an error message. >> >> elfring@Sonne:~/Projekte/Coccinelle/janitor> spatch --parse-cocci >> show_questionable_brelse_usage6.cocci >> … >> minus: parse error: >> File "show_questionable_brelse_usage6.cocci", line 6, column 0, charpos = >> 67 >> around = '', >> whole content = > > That is quite normal. One statement should be followed by another statement. I hope that the support will become better also for assignment expressions. Regards, Markus ___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci
Re: [Cocci] Checking software behaviour according to selected SmPL code variants
On Tue, 9 Jun 2020, Markus Elfring wrote: > Hello, > > My software development attention was caught also by a recent patch. > https://lore.kernel.org/linux-fsdevel/20200608141629.GA1912173@mwanda/ > https://lore.kernel.org/patchwork/patch/1253499/ > > Thus I have tried another tiny script out for the semantic patch language > (according to the software combination “Coccinelle 1.0.8-00104-ge06b9156”). > > > @display@ > expression e, x, y; > @@ > *brelse(e); > *y = x(<+... e ...+>, ...); > > > An usable output is generated then as expected for a test source file > like the following. > https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/fs/exfat/nls.c?id=b676fdbcf4c8424f3c02ed7f31576d99b963bded#n652 > > // SPDX-License-Identifier: GPL-2.0-or-later > // deleted part > static int exfat_load_upcase_table(struct super_block *sb, > sector_t sector, unsigned long long num_sectors, > unsigned int utbl_checksum) > { > struct exfat_sb_info *sbi = EXFAT_SB(sb); > unsigned int sect_size = sb->s_blocksize; > unsigned int i, index = 0; > u32 chksum = 0; > // deleted part > while (sector < num_sectors) { > struct buffer_head *bh; > > bh = sb_bread(sb, sector); > // deleted part > brelse(bh); > chksum = exfat_calc_chksum32(bh->b_data, i, chksum, CS_DEFAULT); > } > // deleted part > } > // deleted part > > > If I omit the specification “, ...” from the function call parameters > because I could be unsure about the number of arguments in other > software situations, I do not get the desired test output as before. This has been discussed before. When you put <+... ...+> in an argument list, it doesn't mean an unknown number of arguments, it means one arguemnt that has something as a subexpression. > > If I omit even the semicolon from the assignment statement in the > search pattern, I get an error message. > > elfring@Sonne:~/Projekte/Coccinelle/janitor> spatch --parse-cocci > show_questionable_brelse_usage6.cocci > … > minus: parse error: > File "show_questionable_brelse_usage6.cocci", line 6, column 0, charpos = 67 > around = '', > whole content = That is quite normal. One statement should be followed by another statement. > > Will such observations influence subsequent software evolution? No. julia___ Cocci mailing list Cocci@systeme.lip6.fr https://systeme.lip6.fr/mailman/listinfo/cocci