Messages in this thread | | | Date | Mon, 30 Jun 2008 20:20:38 +0200 | From | Ingo Molnar <> | Subject | Re: [git pull] core kernel fixes |
| |
* Vegard Nossum <vegard.nossum@gmail.com> wrote:
> On Mon, Jun 30, 2008 at 5:32 PM, Ingo Molnar <mingo@elte.hu> wrote: > > diff --git a/lib/debugobjects.c b/lib/debugobjects.c > > index a76a5e1..85b18d7 100644 > > --- a/lib/debugobjects.c > > +++ b/lib/debugobjects.c > > @@ -68,6 +68,7 @@ static int fill_pool(void) > > { > > gfp_t gfp = GFP_ATOMIC | __GFP_NORETRY | __GFP_NOWARN; > > struct debug_obj *new; > > + unsigned long flags; > > > > if (likely(obj_pool_free >= ODEBUG_POOL_MIN_LEVEL)) > > return obj_pool_free; > > @@ -81,10 +82,10 @@ static int fill_pool(void) > > if (!new) > > return obj_pool_free; > > > > - spin_lock(&pool_lock); > > + spin_lock_irqsave(&pool_lock, flags); > > hlist_add_head(&new->node, &obj_pool); > > obj_pool_free++; > > - spin_unlock(&pool_lock); > > + spin_unlock_irqrestore(&pool_lock, flags); > > } > > return obj_pool_free; > > } > > @@ -110,16 +111,13 @@ static struct debug_obj *lookup_object(void *addr, struct debug_bucket *b) > > } > > > > /* > > - * Allocate a new object. If the pool is empty and no refill possible, > > - * switch off the debugger. > > + * Allocate a new object. If the pool is empty, switch off the debugger. > > */ > > static struct debug_obj * > > alloc_object(void *addr, struct debug_bucket *b, struct debug_obj_descr *descr) > > { > > struct debug_obj *obj = NULL; > > - int retry = 0; > > > > -repeat: > > spin_lock(&pool_lock); > > if (obj_pool.first) { > > obj = hlist_entry(obj_pool.first, typeof(*obj), node); > > @@ -141,9 +139,6 @@ repeat: > > } > > spin_unlock(&pool_lock); > > > > - if (fill_pool() && !obj && !retry++) > > - goto repeat; > > - > > return obj; > > } > > > > @@ -261,6 +256,8 @@ __debug_object_init(void *addr, struct debug_obj_descr *descr, int onstack) > > struct debug_obj *obj; > > unsigned long flags; > > > > + fill_pool(); > > + > > db = get_bucket((unsigned long) addr); > > > > spin_lock_irqsave(&db->lock, flags); > > -- > > Hm. I have to wonder where this patch came from. > > This was my (faulty) patch: http://lkml.org/lkml/2008/6/14/193 > and Daniel J Blueman followed up with this: http://lkml.org/lkml/2008/6/15/27 > > ..but this one looks different from both. I am guessing the last bits > were added (or removed?) by Thomas? > > I am wondering if the final patch was tested with the reproducible > test case (if so, by whom?) and whether should be credited to Daniel > (or Thomas?) instead...
You can use "git log -1 -p --pretty=fuller 50db04dd9c" to see the exact details of the commit:
---------- | commit 50db04dd9c74178e68a981a7127c37252ffb3242 | Author: Vegard Nossum <vegard.nossum@gmail.com> | AuthorDate: Sun Jun 15 00:47:36 2008 +0200 | Commit: Thomas Gleixner <tglx@linutronix.de> | CommitDate: Wed Jun 18 11:09:54 2008 +0200 | | [...] | | [ daniel.blueman@gmail.com: pool_lock needs to be taken irq safe in fill_pool ] | | Reported-by: Daniel J Blueman <daniel.blueman@gmail.com> | Signed-off-by: Vegard Nossum <vegard.nossum@gmail.com> | Signed-off-by: Thomas Gleixner <tglx@linutronix.de> ----------
As you can see it from the Commit line, it was committed by Thomas.
The "[ daniel.blueman: pool_lock ... ]" line shows that Thomas - instead of creating two commits - merged the two fixes into a single commit and credited Daniel for the irq-safe fix. This is the standard technique to squash small patches and to make fixes multi-authored.
The patch was tested with our standard tests so it's certainly good in practice - but i havent specifically tried your testcase (maybe Thomas has). Can you see any problem with the fix?
Ingo
| |