On Fri 12-10-18 10:49:37, Dan Williams wrote:
> devm_memremap_pages() is a facility that can create struct page entries
> for any arbitrary range and give drivers the ability to subvert core
> aspects of page management.
>
> Specifically the facility is tightly integrated with the kernel's memory
> hotplug functionality. It injects an altmap argument deep into the
> architecture specific vmemmap implementation to allow allocating from
> specific reserved pages, and it has Linux specific assumptions about
> page structure reference counting relative to get_user_pages() and
> get_user_pages_fast(). It was an oversight and a mistake that this was
> not marked EXPORT_SYMBOL_GPL from the outset.
One thing is still not clear to me. Both devm_memremap_* and
hmm_devmem_add essentially do the same thing AFAICS. They both allow to
hotplug a device memory. Both rely on the hotplug code (namely
add_pages) which itself is not exported to modules. One is GPL only
while the later is a general export. Is this mismatch desirable?
API exported by the core hotplug is ad-hoc to say the least. Symbols
that we actually export are GPL mostly (only try_offline_node is
EXPORT_SYMBOL without any explanation whatsoever). So I would call it a
general mess tweaked for specific usecases.
I personally do not care about EXPORT_SYMBOL vs. EXPORT_SYMBOL_GPL
much to be honest. I understand an argument that we do not care about
out-of-tree modules a wee bit. I would just be worried those will find a
way around and my experience tells me that it would be much uglier than
what the core kernel can provide. But this seems more political than
technical discussion.
> Again, devm_memremap_pagex() exposes and relies upon core kernel
> internal assumptions and will continue to evolve along with 'struct
> page', memory hotplug, and support for new memory types / topologies.
> Only an in-kernel GPL-only driver is expected to keep up with this
> ongoing evolution. This interface, and functionality derived from this
> interface, is not suitable for kernel-external drivers.
I do not follow this line of argumentation though. We generally do not
care about out-of-tree modules and breaking them if the interface has to
be updated. Also what about GPL out of tree modules?
That being said, I do not mind this patch. You and Christoph are the
authors and therefore it is you to decide. I just find the current
situation confusing to say the least.
> Cc: Michal Hocko <mhocko@xxxxxxxx>
> Cc: "Jérôme Glisse" <jglisse@xxxxxxxxxx>
> Reviewed-by: Christoph Hellwig <hch@xxxxxx>
> Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
> ---
> kernel/memremap.c | 2 +-
> tools/testing/nvdimm/test/iomap.c | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/memremap.c b/kernel/memremap.c
> index 6ec81e0d7a33..1bbb2e892941 100644
> --- a/kernel/memremap.c
> +++ b/kernel/memremap.c
> @@ -232,7 +232,7 @@ void *devm_memremap_pages(struct device *dev, struct dev_pagemap *pgmap)
> err_array:
> return ERR_PTR(error);
> }
> -EXPORT_SYMBOL(devm_memremap_pages);
> +EXPORT_SYMBOL_GPL(devm_memremap_pages);
>
> unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
> {
> diff --git a/tools/testing/nvdimm/test/iomap.c b/tools/testing/nvdimm/test/iomap.c
> index ff9d3a5825e1..ed18a0cbc0c8 100644
> --- a/tools/testing/nvdimm/test/iomap.c
> +++ b/tools/testing/nvdimm/test/iomap.c
> @@ -113,7 +113,7 @@ void *__wrap_devm_memremap_pages(struct device *dev, struct dev_pagemap *pgmap)
> return nfit_res->buf + offset - nfit_res->res.start;
> return devm_memremap_pages(dev, pgmap);
> }
> -EXPORT_SYMBOL(__wrap_devm_memremap_pages);
> +EXPORT_SYMBOL_GPL(__wrap_devm_memremap_pages);
>
> pfn_t __wrap_phys_to_pfn_t(phys_addr_t addr, unsigned long flags)
> {
>
--
Michal Hocko
SUSE Labs
