2020-07-10 12:21:33

by Tian Tao

[permalink] [raw]
Subject: [PATCH v3] vmalloc: Add the right hint when vmalloc failed

In fact "vmalloc=<size>" cmdline option is not available on many
platforms.When the user encounters this error, add the correct
hint to prevent misleading.

Signed-off-by: Tian Tao <[email protected]>

v2:
Add appropriate hints and let users decide if they can increase
the size of the vmalloc by vmalloc=<size> depending on their platform

v3:
change the patch name and commit messages.
---
mm/vmalloc.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index 89e83d3..c6ae7e6 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1238,8 +1238,8 @@ static struct vmap_area *alloc_vmap_area(unsigned long size,
}

if (!(gfp_mask & __GFP_NOWARN) && printk_ratelimit())
- pr_warn("vmap allocation for size %lu failed: use vmalloc=<size> to increase size\n",
- size);
+ pr_warn("vmap allocation for size %lu failed: use vmalloc=<size> to increase size,
+ if your ARCH supports it\n", size);

kmem_cache_free(vmap_area_cachep, va);
return ERR_PTR(-EBUSY);
--
2.7.4


2020-07-10 12:45:21

by Matthew Wilcox

[permalink] [raw]
Subject: Re: [PATCH v3] vmalloc: Add the right hint when vmalloc failed

On Fri, Jul 10, 2020 at 08:17:52PM +0800, Tian Tao wrote:
> In fact "vmalloc=<size>" cmdline option is not available on many
> platforms.When the user encounters this error, add the correct
> hint to prevent misleading.

i don't think this is an improvement.

2020-07-10 13:50:30

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3] vmalloc: Add the right hint when vmalloc failed

> On Fri, Jul 10, 2020 at 08:17:52PM +0800, Tian Tao wrote:
> > In fact "vmalloc=<size>" cmdline option is not available on many
> > platforms.When the user encounters this error, add the correct
> > hint to prevent misleading.
>
> i don't think this is an improvement.
>
Matthew, maybe it should not be considered as improvements?

It is rather an extra information that indicates that your
ARCH may not support "vmalloc" early parameter.

Thanks!

--
Vlad Rezki

2020-07-10 13:53:57

by Matthew Wilcox

[permalink] [raw]
Subject: Re: [PATCH v3] vmalloc: Add the right hint when vmalloc failed

On Fri, Jul 10, 2020 at 03:49:56PM +0200, Uladzislau Rezki wrote:
> > On Fri, Jul 10, 2020 at 08:17:52PM +0800, Tian Tao wrote:
> > > In fact "vmalloc=<size>" cmdline option is not available on many
> > > platforms.When the user encounters this error, add the correct
> > > hint to prevent misleading.
> >
> > i don't think this is an improvement.
> >
> Matthew, maybe it should not be considered as improvements?

Every patch should be an improvement along some dimension. This doesn't
feel like an improvement in any way.

> It is rather an extra information that indicates that your
> ARCH may not support "vmalloc" early parameter.

If anything needs to be improved, it's the documentation, not this
error message.

2020-07-10 14:05:29

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3] vmalloc: Add the right hint when vmalloc failed

On Fri, Jul 10, 2020 at 02:53:01PM +0100, Matthew Wilcox wrote:
> On Fri, Jul 10, 2020 at 03:49:56PM +0200, Uladzislau Rezki wrote:
> > > On Fri, Jul 10, 2020 at 08:17:52PM +0800, Tian Tao wrote:
> > > > In fact "vmalloc=<size>" cmdline option is not available on many
> > > > platforms.When the user encounters this error, add the correct
> > > > hint to prevent misleading.
> > >
> > > i don't think this is an improvement.
> > >
> > Matthew, maybe it should not be considered as improvements?
>
> Every patch should be an improvement along some dimension. This doesn't
> feel like an improvement in any way.
>
> > It is rather an extra information that indicates that your
> > ARCH may not support "vmalloc" early parameter.
>
> If anything needs to be improved, it's the documentation, not this
> error message.
>
As i mentioned before, i do not have a strong opinion about this
patch. From the other hand improving that message looks normal
to me. Especially in a context that it took a time to understand
that ARM64 does not support it, so we have had a precedent. I am
talking about author of this patch.

--
Vlad Rezki