| Message ID | 1464712565-14857-2-git-send-email-lvivier@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 05/31/2016 10:35 AM, Laurent Vivier wrote: > sample from http://coccinellery.org/ > > Signed-off-by: Laurent Vivier <lvivier@redhat.com> > --- > scripts/coccinelle/round.cocci | 19 +++++++++++++++++++ > 1 file changed, 19 insertions(+) > create mode 100644 scripts/coccinelle/round.cocci > > diff --git a/scripts/coccinelle/round.cocci b/scripts/coccinelle/round.cocci > new file mode 100644 > index 0000000..ed06773 > --- /dev/null > +++ b/scripts/coccinelle/round.cocci > @@ -0,0 +1,19 @@ > +// Use macro DIV_ROUND_UP instead of (((n) + (d) - 1) /(d)) > +@@ > +expression e1; > +expression e2; > +@@ > +( > +- ((e1) + e2 - 1) / (e2) > ++ DIV_ROUND_UP(e1,e2) Should this also cover things like: ((e1) & e2) / (e2 + 1) or with bit-shifts in place of division? I don't know how much coccinelle can spot other issues (at least until I review the rest of your series), but at any rate your patch looks like a good first start. I also like the fact that we are committing the script into the repo, so that we can reuse it to catch future additions of the open-coded forms. Reviewed-by: Eric Blake <eblake@redhat.com>
On 31/05/2016 19:12, Eric Blake wrote: > On 05/31/2016 10:35 AM, Laurent Vivier wrote: >> sample from http://coccinellery.org/ >> >> Signed-off-by: Laurent Vivier <lvivier@redhat.com> >> --- >> scripts/coccinelle/round.cocci | 19 +++++++++++++++++++ >> 1 file changed, 19 insertions(+) >> create mode 100644 scripts/coccinelle/round.cocci >> >> diff --git a/scripts/coccinelle/round.cocci b/scripts/coccinelle/round.cocci >> new file mode 100644 >> index 0000000..ed06773 >> --- /dev/null >> +++ b/scripts/coccinelle/round.cocci >> @@ -0,0 +1,19 @@ >> +// Use macro DIV_ROUND_UP instead of (((n) + (d) - 1) /(d)) >> +@@ >> +expression e1; >> +expression e2; >> +@@ >> +( >> +- ((e1) + e2 - 1) / (e2) >> ++ DIV_ROUND_UP(e1,e2) > > Should this also cover things like: > > ((e1) & e2) / (e2 + 1) It's true only if (e2 + 1) is a power of two... and I don't know if we can check that with coccinelle... perhaps we can filter them manually then. > or with bit-shifts in place of division? I don't know how much > coccinelle can spot other issues (at least until I review the rest of > your series), but at any rate your patch looks like a good first start. > I also like the fact that we are committing the script into the repo, so > that we can reuse it to catch future additions of the open-coded forms. > > Reviewed-by: Eric Blake <eblake@redhat.com> > Thanks, Laurent
diff --git a/scripts/coccinelle/round.cocci b/scripts/coccinelle/round.cocci new file mode 100644 index 0000000..ed06773 --- /dev/null +++ b/scripts/coccinelle/round.cocci @@ -0,0 +1,19 @@ +// Use macro DIV_ROUND_UP instead of (((n) + (d) - 1) /(d)) +@@ +expression e1; +expression e2; +@@ +( +- ((e1) + e2 - 1) / (e2) ++ DIV_ROUND_UP(e1,e2) +| +- ((e1) + (e2 - 1)) / (e2) ++ DIV_ROUND_UP(e1,e2) +) + +@@ +expression e1; +expression e2; +@@ +-(DIV_ROUND_UP(e1,e2)) ++DIV_ROUND_UP(e1,e2)
sample from http://coccinellery.org/ Signed-off-by: Laurent Vivier <lvivier@redhat.com> --- scripts/coccinelle/round.cocci | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) create mode 100644 scripts/coccinelle/round.cocci