[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <4a09c32167528e0082559bc170699765f0c49c58.camel@redhat.com>
Date: Mon, 11 Sep 2023 16:56:29 +0200
From: Philipp Stanner <pstanner@...hat.com>
To: Matthew Wilcox <willy@...radead.org>
Cc: linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3] xarray: Document necessary flag in alloc-functions
Oh – well, nope, that's fine.
I just 'abused' v3 as a RESEND as I didn't receive a "merged" message
;)
Let's leave it as it is, thx for merging :)
P.
On Mon, 2023-09-11 at 15:51 +0100, Matthew Wilcox wrote:
> On Mon, Sep 11, 2023 at 04:48:37PM +0200, Philipp Stanner wrote:
> > Calling functions that wrap __xa_alloc() or __xa_alloc_cyclic()
> > without
> > the xarray previously having been initialized with the flag
> > XA_FLAGS_ALLOC being set in xa_init_flags() results in undefined
> > behavior.
> >
> > Document the necessity of setting this flag in all docstrings of
> > functions that wrap said two functions.
> >
> > Signed-off-by: Philipp Stanner <pstanner@...hat.com>
> > ---
> > I used the time available until we can get this merged to create a
> > version-3, improving a few things.
>
> Umm, too late, v2 went upstream last week during the merge window.
>
> Do you still want to change the wording?
>
> > Changes since v2:
> > - Phrase the comment differently: say "requires [...] an xarray
> > [...]"
> > instead of "must be operated on".
> > - Improve the commit message and use the canonical format: a)
> > describe
> > the problem, b) name the solution in imperative form.
> >
> > Regards,
> > P.
> > ---
> > include/linux/xarray.h | 18 ++++++++++++++++++
> > lib/xarray.c | 6 ++++++
> > 2 files changed, 24 insertions(+)
> >
> > diff --git a/include/linux/xarray.h b/include/linux/xarray.h
> > index 741703b45f61..746a17b64aa6 100644
> > --- a/include/linux/xarray.h
> > +++ b/include/linux/xarray.h
> > @@ -856,6 +856,9 @@ static inline int __must_check
> > xa_insert_irq(struct xarray *xa,
> > * stores the index into the @id pointer, then stores the entry at
> > * that index. A concurrent lookup will not see an uninitialised
> > @id.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Any context. Takes and releases the xa_lock. May
> > sleep if
> > * the @gfp flags permit.
> > * Return: 0 on success, -ENOMEM if memory could not be allocated
> > or
> > @@ -886,6 +889,9 @@ static inline __must_check int xa_alloc(struct
> > xarray *xa, u32 *id,
> > * stores the index into the @id pointer, then stores the entry at
> > * that index. A concurrent lookup will not see an uninitialised
> > @id.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Any context. Takes and releases the xa_lock while
> > * disabling softirqs. May sleep if the @gfp flags permit.
> > * Return: 0 on success, -ENOMEM if memory could not be allocated
> > or
> > @@ -916,6 +922,9 @@ static inline int __must_check
> > xa_alloc_bh(struct xarray *xa, u32 *id,
> > * stores the index into the @id pointer, then stores the entry at
> > * that index. A concurrent lookup will not see an uninitialised
> > @id.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Process context. Takes and releases the xa_lock while
> > * disabling interrupts. May sleep if the @gfp flags permit.
> > * Return: 0 on success, -ENOMEM if memory could not be allocated
> > or
> > @@ -949,6 +958,9 @@ static inline int __must_check
> > xa_alloc_irq(struct xarray *xa, u32 *id,
> > * The search for an empty entry will start at @next and will wrap
> > * around if necessary.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Any context. Takes and releases the xa_lock. May
> > sleep if
> > * the @gfp flags permit.
> > * Return: 0 if the allocation succeeded without wrapping. 1 if
> > the
> > @@ -983,6 +995,9 @@ static inline int xa_alloc_cyclic(struct xarray
> > *xa, u32 *id, void *entry,
> > * The search for an empty entry will start at @next and will wrap
> > * around if necessary.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Any context. Takes and releases the xa_lock while
> > * disabling softirqs. May sleep if the @gfp flags permit.
> > * Return: 0 if the allocation succeeded without wrapping. 1 if
> > the
> > @@ -1017,6 +1032,9 @@ static inline int xa_alloc_cyclic_bh(struct
> > xarray *xa, u32 *id, void *entry,
> > * The search for an empty entry will start at @next and will wrap
> > * around if necessary.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Process context. Takes and releases the xa_lock while
> > * disabling interrupts. May sleep if the @gfp flags permit.
> > * Return: 0 if the allocation succeeded without wrapping. 1 if
> > the
> > diff --git a/lib/xarray.c b/lib/xarray.c
> > index 2071a3718f4e..2b07c332d26b 100644
> > --- a/lib/xarray.c
> > +++ b/lib/xarray.c
> > @@ -1802,6 +1802,9 @@ EXPORT_SYMBOL(xa_get_order);
> > * stores the index into the @id pointer, then stores the entry at
> > * that index. A concurrent lookup will not see an uninitialised
> > @id.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Any context. Expects xa_lock to be held on entry.
> > May
> > * release and reacquire xa_lock if @gfp flags permit.
> > * Return: 0 on success, -ENOMEM if memory could not be allocated
> > or
> > @@ -1850,6 +1853,9 @@ EXPORT_SYMBOL(__xa_alloc);
> > * The search for an empty entry will start at @next and will wrap
> > * around if necessary.
> > *
> > + * Requires the xarray to be initialized with flag XA_FLAGS_ALLOC
> > set
> > + * in xa_init_flags().
> > + *
> > * Context: Any context. Expects xa_lock to be held on entry.
> > May
> > * release and reacquire xa_lock if @gfp flags permit.
> > * Return: 0 if the allocation succeeded without wrapping. 1 if
> > the
> > --
> > 2.41.0
> >
>
Powered by blists - more mailing lists