| Message ID | 20231025-ethtool_puts_impl-v1-3-6a53a93d3b72@google.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | ethtool: Add ethtool_puts() | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Guessing tree name failed - patch did not apply |
On Wed, 2023-10-25 at 23:40 +0000, Justin Stitt wrote: > Add some warnings for using ethtool_sprintf() where a simple > ethtool_puts() would suffice. > > The two cases are: > > 1) Use ethtool_sprintf() with just two arguments: > > ethtool_sprintf(&data, driver[i].name); OK. > or > 2) Use ethtool_sprintf() with a standalone "%s" fmt string: > > ethtool_sprintf(&data, "%s", driver[i].name); I'm rather doubt this is really desired or appropriate.
On 10/26/23 01:40, Justin Stitt wrote: > Add some warnings for using ethtool_sprintf() where a simple > ethtool_puts() would suffice. > > The two cases are: > > 1) Use ethtool_sprintf() with just two arguments: > | ethtool_sprintf(&data, driver[i].name); > or > 2) Use ethtool_sprintf() with a standalone "%s" fmt string: > | ethtool_sprintf(&data, "%s", driver[i].name); > > The former may cause -Wformat-security warnings while the latter is just > not preferred. Both are safely in the category of warnings, not errors. > > Signed-off-by: Justin Stitt <justinstitt@google.com> > --- > scripts/checkpatch.pl | 13 +++++++++++++ > 1 file changed, 13 insertions(+) > > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl > index 7d16f863edf1..1ba9ce778746 100755 > --- a/scripts/checkpatch.pl > +++ b/scripts/checkpatch.pl > @@ -7020,6 +7020,19 @@ sub process { > "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr); > } > > +# ethtool_sprintf uses that should likely be ethtool_puts > + if ( $line =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\)/ ) { no need for whitespace right after opening parenthesis, same at the end Does it work for ethtool_sprintf(calls broken into multiple lines)? BTW, I really like this series! > + WARN("ETHTOOL_SPRINTF", > + "Prefer ethtool_puts over ethtool_sprintf with only two arguments" . $herecurr); > + } > + > + # use $rawline because $line loses %s via sanitization and thus we can't match against it. > + if ( $rawline =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*\"\%s\"\s*,\s*$FuncArg\s*\)/ ) { > + WARN("ETHTOOL_SPRINTF2", > + "Prefer ethtool_puts over ethtool_sprintf with standalone \"%s\" specifier" . $herecurr); > + } > + > + > # typecasts on min/max could be min_t/max_t > if ($perl_version_ok && > defined $stat && >
On Wed, 2023-10-25 at 23:40 +0000, Justin Stitt wrote: > Add some warnings for using ethtool_sprintf() where a simple > ethtool_puts() would suffice. Hi again Justin. After I read patch 1/3 I don't object at all. spatch/cocci will always be a better option than checkpatch for conversions like this because it's a proper grammar parser and checkpatch is a stupid little perl script. If you resubmit this please: > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl [] > @@ -7020,6 +7020,19 @@ sub process { > "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr); > } > > +# ethtool_sprintf uses that should likely be ethtool_puts > + if ( $line =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\)/ ) { > + WARN("ETHTOOL_SPRINTF", > + "Prefer ethtool_puts over ethtool_sprintf with only two arguments" . $herecurr); > + } > + > + # use $rawline because $line loses %s via sanitization and thus we can't match against it. > + if ( $rawline =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*\"\%s\"\s*,\s*$FuncArg\s*\)/ ) { > + WARN("ETHTOOL_SPRINTF2", > + "Prefer ethtool_puts over ethtool_sprintf with standalone \"%s\" specifier" . $herecurr); > + } o remove the whitespace before and after the parentheses o use the same type "ETHTOOL_SPRINTF" or maybe "PREFER_ETHTOOL_PUTS" for both warnings. o Add a newline on the message output o Add a --fix option Something like: --- scripts/checkpatch.pl | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl index 25fdb7fda1128..6924731110d87 100755 --- a/scripts/checkpatch.pl +++ b/scripts/checkpatch.pl @@ -7011,6 +7011,25 @@ sub process { "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr); } +# ethtool_sprintf uses that should likely be ethtool_puts + if ($line =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\)/) { + if (WARN("PREFER_ETHTOOL_PUTS", + "Prefer ethtool_puts over ethtool_sprintf with only two arguments\n" . $herecurr) && + $fix) { + $fixed[$fixlinenr] =~ s/\bethtool_sprintf\s*\(\s*($FuncArg)\s*,\s*($FuncArg)/ethtool_puts($1, $7)/; + } + } + + # use $rawline because $line loses %s via sanitization and thus we can't match against it. + if ($rawline =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*\"\%s\"\s*,\s*$FuncArg\s*\)/) { + if (WARN("PREFER_ETHTOOL_PUTS", + "Prefer ethtool_puts over ethtool_sprintf with standalone \"%s\" specifier\n" . $herecurr) && + $fix) { + $fixed[$fixlinenr] =~ s/\bethtool_sprintf\s*\(\s*($FuncArg)\s*,\s*"\%s"\s*,\s*($FuncArg)/ethtool_puts($1, $7)/; + } + } + + # typecasts on min/max could be min_t/max_t if ($perl_version_ok && defined $stat &&
diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl index 7d16f863edf1..1ba9ce778746 100755 --- a/scripts/checkpatch.pl +++ b/scripts/checkpatch.pl @@ -7020,6 +7020,19 @@ sub process { "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr); } +# ethtool_sprintf uses that should likely be ethtool_puts + if ( $line =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\)/ ) { + WARN("ETHTOOL_SPRINTF", + "Prefer ethtool_puts over ethtool_sprintf with only two arguments" . $herecurr); + } + + # use $rawline because $line loses %s via sanitization and thus we can't match against it. + if ( $rawline =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*\"\%s\"\s*,\s*$FuncArg\s*\)/ ) { + WARN("ETHTOOL_SPRINTF2", + "Prefer ethtool_puts over ethtool_sprintf with standalone \"%s\" specifier" . $herecurr); + } + + # typecasts on min/max could be min_t/max_t if ($perl_version_ok && defined $stat &&
Add some warnings for using ethtool_sprintf() where a simple ethtool_puts() would suffice. The two cases are: 1) Use ethtool_sprintf() with just two arguments: | ethtool_sprintf(&data, driver[i].name); or 2) Use ethtool_sprintf() with a standalone "%s" fmt string: | ethtool_sprintf(&data, "%s", driver[i].name); The former may cause -Wformat-security warnings while the latter is just not preferred. Both are safely in the category of warnings, not errors. Signed-off-by: Justin Stitt <justinstitt@google.com> --- scripts/checkpatch.pl | 13 +++++++++++++ 1 file changed, 13 insertions(+)