void arch_mm_init(unsigned long *pt_pfn,
unsigned long pt_base, unsigned long nr_pt_frames, unsigned long nr_pages)
{
unsigned long start_pfn = (nr_pt_frames - NOT_L1_FRAMES) * L1_PAGETABLE_ENTRIES;
unsigned long end_pfn = nr_pages;
build_page_tables((pgentry_t *)pt_base,
pt_pfn, pfn_to_virt(start_pfn), pfn_to_virt(end_pfn));
}
/*ARGSUSED*/
void
startup_kernel(void)
{
char *cmdline;
uintptr_t addr;
#if defined(__xpv)
physdev_set_iopl_t set_iopl;
#endif /* __xpv */
/*
* At this point we are executing in a 32 bit real mode.
*/
#if defined(__xpv)
cmdline = (char *)xen_info->cmd_line;
#else /* __xpv */
cmdline = (char *)mb_info->cmdline;
#endif /* __xpv */
prom_debug = (strstr(cmdline, "prom_debug") != NULL);
map_debug = (strstr(cmdline, "map_debug") != NULL);
#if defined(__xpv)
/*
* For dom0, before we initialize the console subsystem we'll
* need to enable io operations, so set I/O priveldge level to 1.
*/
if (DOMAIN_IS_INITDOMAIN(xen_info)) {
set_iopl.iopl = 1;
(void) HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
}
#endif /* __xpv */
bcons_init(cmdline);
DBG_MSG("\n\nSolaris prekernel set: ");
DBG_MSG(cmdline);
DBG_MSG("\n");
if (strstr(cmdline, "multiboot") != NULL) {
dboot_panic(NO_MULTIBOOT);
}
/*
* boot info must be 16 byte aligned for 64 bit kernel ABI
*/
addr = (uintptr_t)boot_info;
addr = (addr + 0xf) & ~0xf;
bi = (struct xboot_info *)addr;
DBG((uintptr_t)bi);
bi->bi_cmdline = (native_ptr_t)(uintptr_t)cmdline;
/*
* Need correct target_kernel_text value
*/
#if defined(_BOOT_TARGET_amd64)
target_kernel_text = KERNEL_TEXT_amd64;
#elif defined(__xpv)
target_kernel_text = KERNEL_TEXT_i386_xpv;
#else
target_kernel_text = KERNEL_TEXT_i386;
#endif
DBG(target_kernel_text);
#if defined(__xpv)
/*
* XXPV Derive this stuff from CPUID / what the hypervisor has enabled
*/
#if defined(_BOOT_TARGET_amd64)
/*
* 64-bit hypervisor.
*/
amd64_support = 1;
pae_support = 1;
#else /* _BOOT_TARGET_amd64 */
/*
* See if we are running on a PAE Hypervisor
*/
{
xen_capabilities_info_t caps;
if (HYPERVISOR_xen_version(XENVER_capabilities, &caps) != 0)
dboot_panic("HYPERVISOR_xen_version(caps) failed");
caps[sizeof (caps) - 1] = 0;
if (prom_debug)
dboot_printf("xen capabilities %s\n", caps);
if (strstr(caps, "x86_32p") != NULL)
pae_support = 1;
}
#endif /* _BOOT_TARGET_amd64 */
{
xen_platform_parameters_t p;
if (HYPERVISOR_xen_version(XENVER_platform_parameters, &p) != 0)
dboot_panic("HYPERVISOR_xen_version(parms) failed");
DBG(p.virt_start);
mfn_to_pfn_mapping = (pfn_t *)(xen_virt_start = p.virt_start);
}
/*
* The hypervisor loads stuff starting at 1Gig
*/
mfn_base = ONE_GIG;
DBG(mfn_base);
/*
* enable writable page table mode for the hypervisor
*/
if (HYPERVISOR_vm_assist(VMASST_CMD_enable,
VMASST_TYPE_writable_pagetables) < 0)
dboot_panic("HYPERVISOR_vm_assist(writable_pagetables) failed");
/*
* check for NX support
*/
if (pae_support) {
uint32_t eax = 0x80000000;
uint32_t edx = get_cpuid_edx(&eax);
if (eax >= 0x80000001) {
eax = 0x80000001;
edx = get_cpuid_edx(&eax);
if (edx & CPUID_AMD_EDX_NX)
NX_support = 1;
}
}
#if !defined(_BOOT_TARGET_amd64)
/*
* The 32-bit hypervisor uses segmentation to protect itself from
* guests. This means when a guest attempts to install a flat 4GB
* code or data descriptor the 32-bit hypervisor will protect itself
* by silently shrinking the segment such that if the guest attempts
* any access where the hypervisor lives a #gp fault is generated.
* The problem is that some applications expect a full 4GB flat
* segment for their current thread pointer and will use negative
* offset segment wrap around to access data. TLS support in linux
* brand is one example of this.
*
* The 32-bit hypervisor can catch the #gp fault in these cases
* and emulate the access without passing the #gp fault to the guest
* but only if VMASST_TYPE_4gb_segments is explicitly turned on.
* Seems like this should have been the default.
* Either way, we want the hypervisor -- and not Solaris -- to deal
* to deal with emulating these accesses.
*/
if (HYPERVISOR_vm_assist(VMASST_CMD_enable,
VMASST_TYPE_4gb_segments) < 0)
dboot_panic("HYPERVISOR_vm_assist(4gb_segments) failed");
#endif /* !_BOOT_TARGET_amd64 */
#else /* __xpv */
/*
* use cpuid to enable MMU features
*/
if (have_cpuid()) {
uint32_t eax, edx;
eax = 1;
edx = get_cpuid_edx(&eax);
if (edx & CPUID_INTC_EDX_PSE)
largepage_support = 1;
if (edx & CPUID_INTC_EDX_PGE)
pge_support = 1;
if (edx & CPUID_INTC_EDX_PAE)
pae_support = 1;
eax = 0x80000000;
edx = get_cpuid_edx(&eax);
if (eax >= 0x80000001) {
eax = 0x80000001;
edx = get_cpuid_edx(&eax);
if (edx & CPUID_AMD_EDX_LM)
amd64_support = 1;
if (edx & CPUID_AMD_EDX_NX)
NX_support = 1;
}
} else {
dboot_printf("cpuid not supported\n");
}
#endif /* __xpv */
#if defined(_BOOT_TARGET_amd64)
if (amd64_support == 0)
dboot_panic("long mode not supported, rebooting");
else if (pae_support == 0)
dboot_panic("long mode, but no PAE; rebooting");
#else
/*
* Allow the command line to over-ride use of PAE for 32 bit.
*/
if (strstr(cmdline, "disablePAE=true") != NULL) {
pae_support = 0;
NX_support = 0;
amd64_support = 0;
}
#endif
/*
* initialize the simple memory allocator
*/
init_mem_alloc();
#if !defined(__xpv) && !defined(_BOOT_TARGET_amd64)
/*
* disable PAE on 32 bit h/w w/o NX and < 4Gig of memory
*/
if (max_mem < FOUR_GIG && NX_support == 0)
pae_support = 0;
#endif
/*
* configure mmu information
*/
if (pae_support) {
shift_amt = shift_amt_pae;
ptes_per_table = 512;
pte_size = 8;
lpagesize = TWO_MEG;
#if defined(_BOOT_TARGET_amd64)
top_level = 3;
#else
top_level = 2;
#endif
} else {
pae_support = 0;
NX_support = 0;
shift_amt = shift_amt_nopae;
ptes_per_table = 1024;
pte_size = 4;
lpagesize = FOUR_MEG;
top_level = 1;
}
DBG(pge_support);
DBG(NX_support);
DBG(largepage_support);
DBG(amd64_support);
DBG(top_level);
DBG(pte_size);
DBG(ptes_per_table);
DBG(lpagesize);
#if defined(__xpv)
ktext_phys = ONE_GIG; /* from UNIX Mapfile */
#else
ktext_phys = FOUR_MEG; /* from UNIX Mapfile */
#endif
#if !defined(__xpv) && defined(_BOOT_TARGET_amd64)
/*
* For grub, copy kernel bits from the ELF64 file to final place.
*/
DBG_MSG("\nAllocating nucleus pages.\n");
ktext_phys = (uintptr_t)do_mem_alloc(ksize, FOUR_MEG);
if (ktext_phys == 0)
dboot_panic("failed to allocate aligned kernel memory");
if (dboot_elfload64(mb_header.load_addr) != 0)
dboot_panic("failed to parse kernel ELF image, rebooting");
#endif
DBG(ktext_phys);
/*
* Allocate page tables.
*/
build_page_tables();
/*
* return to assembly code to switch to running kernel
*/
entry_addr_low = (uint32_t)target_kernel_text;
DBG(entry_addr_low);
bi->bi_use_largepage = largepage_support;
bi->bi_use_pae = pae_support;
bi->bi_use_pge = pge_support;
bi->bi_use_nx = NX_support;
#if defined(__xpv)
bi->bi_next_paddr = next_avail_addr - mfn_base;
DBG(bi->bi_next_paddr);
bi->bi_next_vaddr = (native_ptr_t)next_avail_addr;
DBG(bi->bi_next_vaddr);
/*
* unmap unused pages in start area to make them available for DMA
*/
while (next_avail_addr < scratch_end) {
(void) HYPERVISOR_update_va_mapping(next_avail_addr,
0, UVMF_INVLPG | UVMF_LOCAL);
next_avail_addr += MMU_PAGESIZE;
}
bi->bi_xen_start_info = (uintptr_t)xen_info;
DBG((uintptr_t)HYPERVISOR_shared_info);
bi->bi_shared_info = (native_ptr_t)HYPERVISOR_shared_info;
bi->bi_top_page_table = (uintptr_t)top_page_table - mfn_base;
#else /* __xpv */
bi->bi_next_paddr = next_avail_addr;
DBG(bi->bi_next_paddr);
bi->bi_next_vaddr = (uintptr_t)next_avail_addr;
DBG(bi->bi_next_vaddr);
bi->bi_mb_info = (uintptr_t)mb_info;
bi->bi_top_page_table = (uintptr_t)top_page_table;
#endif /* __xpv */
bi->bi_kseg_size = FOUR_MEG;
DBG(bi->bi_kseg_size);
#ifndef __xpv
if (map_debug)
dump_tables();
#endif
DBG_MSG("\n\n*** DBOOT DONE -- back to asm to jump to kernel\n\n");
}
