lkml.org 
[lkml]   [2023]   [Aug]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    SubjectRe: [PATCH] tee: amdtee: add support for use of cma region
    From
    On 8/18/23 07:42, Devaraj Rangasamy wrote:
    > --- a/arch/x86/kernel/setup.c
    > +++ b/arch/x86/kernel/setup.c
    > @@ -1223,6 +1223,8 @@ void __init setup_arch(char **cmdline_p)
    > initmem_init();
    > dma_contiguous_reserve(max_pfn_mapped << PAGE_SHIFT);
    >
    > + amdtee_cma_reserve();
    > +
    > if (boot_cpu_has(X86_FEATURE_GBPAGES))
    > hugetlb_cma_reserve(PUD_SHIFT - PAGE_SHIFT);

    Right now, we have *A* global CMA pool set up in
    dma_contiguous_reserve() that everyone shares.

    Why does this *one* driver deserve to be a special snowflake and get its
    own private CMA area and own command-line options?

    It seems to me like you should just tell users to set
    CONFIG_CMA_SIZE_MBYTES or use cma=size on the command-line and just use
    the global pool. If you want to make it a special snowflake, there's a
    much higher bar to clear, and there's zero justification for that right now.

    Oh, and this:

    > static int pool_op_alloc(struct tee_shm_pool *pool, struct tee_shm *shm,
    > size_t size, size_t align)
    > {
    > unsigned int order = get_order(size);
    > unsigned long va;
    > int rc;
    >
    > /*
    > * Ignore alignment since this is already going to be page aligned
    > * and there's no need for any larger alignment.
    > */
    > va = __get_free_pages(GFP_KERNEL | __GFP_ZERO, order);

    is goofy. It either only needs powers-of-2 and can take "order" as an
    argument instead of 'size', or it should be using alloc_pages_exact() to
    avoid wasting memory.

    \
     
     \ /
      Last update: 2023-08-18 17:27    [W:6.398 / U:0.148 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site