Re: [PATCH v1 2/3] kernel/resource: make walk_mem_res() find all busy IORESOURCE_MEM resources

2021-03-23 Thread David Hildenbrand
On 23.03.21 12:08, Andy Shevchenko wrote: On Mon, Mar 22, 2021 at 05:01:59PM +0100, David Hildenbrand wrote: It used to be true that we can have system RAM only on the first level in the resourc tree. However, this is no longer holds for driver-managed system RAM (i.e., dax/kmem and virtio-mem).

Re: [PATCH v1 2/3] kernel/resource: make walk_mem_res() find all busy IORESOURCE_MEM resources

2021-03-23 Thread Andy Shevchenko
On Mon, Mar 22, 2021 at 05:01:59PM +0100, David Hildenbrand wrote: > It used to be true that we can have system RAM only on the first level > in the resourc tree. However, this is no longer holds for driver-managed > system RAM (i.e., dax/kmem and virtio-mem). > > The function walk_mem_res() only

Re: [PATCH v1 2/3] kernel/resource: make walk_mem_res() find all busy IORESOURCE_MEM resources

2021-03-22 Thread Dan Williams
On Mon, Mar 22, 2021 at 9:02 AM David Hildenbrand wrote: > > It used to be true that we can have system RAM only on the first level > in the resourc tree. However, this is no longer holds for driver-managed > system RAM (i.e., dax/kmem and virtio-mem). > > The function walk_mem_res() only consider

[PATCH v1 2/3] kernel/resource: make walk_mem_res() find all busy IORESOURCE_MEM resources

2021-03-22 Thread David Hildenbrand
It used to be true that we can have system RAM only on the first level in the resourc tree. However, this is no longer holds for driver-managed system RAM (i.e., dax/kmem and virtio-mem). The function walk_mem_res() only consideres the first level and is used in arch/x86/mm/ioremap.c:__ioremap_che