On 5/14/25 4:32 PM, Jan Beulich wrote:
On 06.05.2025 18:51, Oleksii Kurochko wrote:@@ -583,3 +584,36 @@ void *__init arch_vmap_virt_end(void) { return (void *)(VMAP_VIRT_START + VMAP_VIRT_SIZE); } + +static void *ioremap_attr(paddr_t start, size_t len, pte_attr_t attributes) +{ + mfn_t mfn = _mfn(PFN_DOWN(start)); + unsigned int offs = start & (PAGE_SIZE - 1); + unsigned int nr = PFN_UP(offs + len); + void *ptr = __vmap(&mfn, nr, 1, 1, attributes, VMAP_DEFAULT); + + if ( ptr == NULL ) + return NULL; + + return ptr + offs; +} + +void __iomem *ioremap_nocache(paddr_t start, size_t len) +{ + return ioremap_attr(start, len, PAGE_HYPERVISOR_NOCACHE); +}Why do you need both this and ...+void __iomem *ioremap_cache(paddr_t start, size_t len) +{ + return ioremap_attr(start, len, PAGE_HYPERVISOR); +} + +void __iomem *ioremap_wc(paddr_t start, size_t len) +{ + return ioremap_attr(start, len, PAGE_HYPERVISOR_WC); +} + +void *ioremap(paddr_t pa, size_t len) +{ + return ioremap_attr(pa, len, PAGE_HYPERVISOR_NOCACHE); +}... this? And why's the 1st parameter named differently for this last one? Can't they all be in sync in this regard?
Originally, I thought|ioremap_nocache()| was needed because it is used in common code. However, I now realize that the calls to|ioremap_nocache()| in|xen/drivers/char| are also ARM-specific. I assume all the UART drivers currently using|ioremap_nocache|() are intended for ARM. I'll drop|ioremap_nocache()| for RISC-V, and I think it makes sense to create a separate patch to clean this up for ARM as well. ~ Oleksii
