The mask calculation in pdx_init_mask is wrong when the first bank starts at address 0x0. The reason is that pdx_init_mask will do '0 - 1' causing an underflow. As a result, the mask becomes 0xffffffffffffffff which is the biggest possible mask and ends up causing a significant memory waste in the frametable size computation.
For instance, on platforms that have a low memory bank starting at 0x0 and a high memory bank, the frametable will end up covering all the holes in between. The purpose of the mask is to be passed as a parameter to pfn_pdx_hole_setup, which based on the mask parameter calculates pfn_pdx_hole_shift, pfn_pdx_bottom_mask, etc. which are actually the important masks for frametable initialization later on. pfn_pdx_hole_setup never compresses addresses below MAX_ORDER bits (1GB on ARM). Thus, it is safe to initialize mask passing 1ULL << (MAX_ORDER + PAGE_SHIFT) as start address to pdx_init_mask. Signed-off-by: Stefano Stabellini <stefa...@xilinx.com> CC: jbeul...@suse.com CC: andrew.coop...@citrix.com CC: george.dun...@eu.citrix.com CC: ian.jack...@eu.citrix.com CC: konrad.w...@oracle.com CC: t...@xen.org CC: wei.l...@citrix.com --- Changes in v3: - improve in-code comments Changes in v2: - update commit message - add in-code comments regarding update sites - improve in-code comments - move the mask initialization changes to pdx_init_mask --- xen/arch/arm/setup.c | 9 ++++++++- xen/common/pdx.c | 8 +++++++- 2 files changed, 15 insertions(+), 2 deletions(-) diff --git a/xen/arch/arm/setup.c b/xen/arch/arm/setup.c index ccb0f181ea..45312df006 100644 --- a/xen/arch/arm/setup.c +++ b/xen/arch/arm/setup.c @@ -482,7 +482,14 @@ static void __init init_pdx(void) { paddr_t bank_start, bank_size, bank_end; - u64 mask = pdx_init_mask(bootinfo.mem.bank[0].start); + /* + * Arm does not have any restrictions on the bits to compress. Pass 0 to + * let the common code further restrict the mask. + * + * If the logic changes in pfn_pdx_hole_setup we might have to + * update this function too. + */ + u64 mask = pdx_init_mask(0x0); int bank; for ( bank = 0 ; bank < bootinfo.mem.nr_banks; bank++ ) diff --git a/xen/common/pdx.c b/xen/common/pdx.c index bb7e437049..a3c6f4c1ee 100644 --- a/xen/common/pdx.c +++ b/xen/common/pdx.c @@ -50,9 +50,12 @@ static u64 __init fill_mask(u64 mask) return mask; } +/* + * We don't compress the first MAX_ORDER bit of the addresses. + */ u64 __init pdx_init_mask(u64 base_addr) { - return fill_mask(base_addr - 1); + return fill_mask(max(base_addr, (u64)1 << (MAX_ORDER + PAGE_SHIFT)) - 1); } u64 __init pdx_region_mask(u64 base, u64 len) @@ -80,6 +83,9 @@ void __init pfn_pdx_hole_setup(unsigned long mask) * This guarantees that page-pointer arithmetic remains valid within * contiguous aligned ranges of 2^MAX_ORDER pages. Among others, our * buddy allocator relies on this assumption. + * + * If the logic changes here, we might have to update the ARM specific + * init_pdx too. */ for ( j = MAX_ORDER-1; ; ) { -- 2.17.1 _______________________________________________ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel