On Sat, Oct 22, 2022 at 07:45:23PM -0400, Elad Lahav wrote: > Fine with me. > > (Of course, tabs are an invention of the Devil, and I felt very > Faustian as I re-formatted the code before submitting the patch. The > fact that I still got the indentation wrong just proves that hell hath > no fury like tabified code.) My feeling is that you can leave out the "tabified" without losing much, but opinions may vary. ;-) > --Elad > > On Sat, 22 Oct 2022 at 11:02, Akira Yokosawa <akiyks@xxxxxxxxx> wrote: > > > > perfbook's coding style mostly follows that of the Linux kernel [1]. > > One exception is the indents of continuation lines, where TABs are > > used as many as the leading line, while white spaces are used for > > deeper level indents in continuation lines. > > > > This is to keep alignment regardless of TAB:SPC setting. > > In two-column layout (2c), code snippets are rendered with > > TAB:SPC == 1:2, while in one-column layouts (1c and eb), they are > > rendered with TAB:SPC == 1:8. > > > > There is another convention in label names for referencing line > > numbers. The base part of such a label often looks like > > "ln:<chapter>:<base name of source file>:<some id in the file>" > > to make it easy to see where the extracted snippet files (.fcv) > > originates. > > > > Stick to these conventions and make other tweaks listed below: > > > > - Extend the snippet so that the definitions of the structs used in > > the code can be presented. > > - Declare the local variable itr at the beginning of > > tree_for_each_rec(). > > - Fold a long line which exceeded the column width of two-column > > layout after the indent adjustment. > > - Mention the source file name where the reference to the snippet > > is made for the first time in the text. > > > > Link: [1] https://www.kernel.org/doc/html/latest/process/coding-style.html > > Signed-off-by: Akira Yokosawa <akiyks@xxxxxxxxx> > > Cc: Elad Lahav <e2lahav@xxxxxxxxx> > > --- > > Hi Paul, > > > > I know I'm doing multiple things at once here. > > If you'd prefer a patch series, I can respin. I took this one as is, thank you very much! Thanx, Paul > > Thanks, Akira > > -- > > CodeSamples/locking/rec_tree_itr.c | 15 +++++++++------ > > locking/locking.tex | 4 ++-- > > 2 files changed, 11 insertions(+), 8 deletions(-) > > > > diff --git a/CodeSamples/locking/rec_tree_itr.c b/CodeSamples/locking/rec_tree_itr.c > > index d091e4cc15c7..20b0d7bc4fce 100644 > > --- a/CodeSamples/locking/rec_tree_itr.c > > +++ b/CodeSamples/locking/rec_tree_itr.c > > @@ -20,6 +20,7 @@ > > > > #include "../api.h" > > > > +//\begin{snippet}[labelbase=ln:locking:rec_tree_itr:tree_for_each,commandchars=\\\@\$] > > struct node { > > int data; > > int nchildren; > > @@ -31,15 +32,16 @@ struct tree { > > struct node *root; > > }; > > > > -//\begin{snippet}[labelbase=ln:locking:rec_tree_iterator:tree_for_each,commandchars=\\\@\$] > > void tree_for_each_rec(struct tree *tr, struct node *nd, > > - void (*callback)(struct node *)) > > + void (*callback)(struct node *)) > > { > > + struct node **itr; > > + > > spin_unlock(&tr->s); > > callback(nd); > > spin_lock(&tr->s); > > > > - struct node **itr = nd->children; > > + itr = nd->children; > > for (int i = 0; i < nd->nchildren; i++) { > > tree_for_each_rec(tr, *itr, callback); > > itr++; > > @@ -47,7 +49,7 @@ void tree_for_each_rec(struct tree *tr, struct node *nd, > > } > > > > void tree_for_each(struct tree *tr, > > - void (*callback)(struct node *)) > > + void (*callback)(struct node *)) > > { > > spin_lock(&tr->s); > > tree_for_each_rec(tr, tr->root, callback); > > @@ -55,12 +57,13 @@ void tree_for_each(struct tree *tr, > > } > > > > void tree_add(struct tree *tr, struct node *parent, > > - struct node *new_child) > > + struct node *new_child) > > { > > spin_lock(&tr->s); > > parent->nchildren++; > > parent->children = realloc(parent->children, > > - sizeof(struct node *) * parent->nchildren); > > + sizeof(struct node *) * > > + parent->nchildren); > > parent->children[parent->nchildren - 1] = new_child; > > spin_unlock(&tr->s); > > } > > diff --git a/locking/locking.tex b/locking/locking.tex > > index c990d4f86a7b..ac8bab8c8a1b 100644 > > --- a/locking/locking.tex > > +++ b/locking/locking.tex > > @@ -383,7 +383,7 @@ Because the \co{qsort()} comparison function rarely acquires locks, > > let's switch to a different example. > > > > Consider the recursive tree iterator in > > -\cref{lst:locking:Recursive Tree Iterator}. > > +\cref{lst:locking:Recursive Tree Iterator} (\path{rec_tree_itr.c}). > > The iterator visits every node in the tree, invoking a user's callback > > function. > > The tree lock is released before the invocation and re-acquired after return. > > @@ -424,7 +424,7 @@ Alternatively, the state can be re-initialized once the lock is > > re-acquired after the callback function returns. > > > > \begin{listing} > > -\input{CodeSamples/locking/rec_tree_iterator@tree_for_each.fcv} > > +\input{CodeSamples/locking/rec_tree_itr@tree_for_each.fcv} > > \caption{Recursive Tree Iterator} > > \label{lst:locking:Recursive Tree Iterator} > > \end{listing} > > > > base-commit: 55c6bfdbfd44cc1090b54dceeb004ae254c93f7c > > -- > > 2.25.1 > >
