| Message ID | 1473711511-11931-6-git-send-email-daniel.kiper@oracle.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
>>> On 12.09.16 at 22:18, <daniel.kiper@oracle.com> wrote: > --- a/xen/arch/x86/domain_page.c > +++ b/xen/arch/x86/domain_page.c > @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) > * domain's page tables but current may point at another domain's VCPU. > * Return NULL as though current is not properly set up yet. > */ > - if ( efi_enabled && efi_rs_using_pgtables() ) > + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) I think the efi_enabled() here is pointless now. With this dropped (unless you know of a reason that it needs to stay) Reviewed-by: Jan Beulich <jbeulich@suse.com> Jan
On Mon, Sep 19, 2016 at 05:58:46AM -0600, Jan Beulich wrote: > >>> On 12.09.16 at 22:18, <daniel.kiper@oracle.com> wrote: > > --- a/xen/arch/x86/domain_page.c > > +++ b/xen/arch/x86/domain_page.c > > @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) > > * domain's page tables but current may point at another domain's VCPU. > > * Return NULL as though current is not properly set up yet. > > */ > > - if ( efi_enabled && efi_rs_using_pgtables() ) > > + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) > > I think the efi_enabled() here is pointless now. Nope, it seems that Xen will blow up on BUG() in xen/arch/x86/efi/stub.c:efi_rs_using_pgtables() if compiler/linker cannot be used to build proper PE binary. Of course we can change efi_rs_using_pgtables() to return false in such case. > With this dropped (unless you know of a reason that it needs to stay) > Reviewed-by: Jan Beulich <jbeulich@suse.com> Thanks a lot! By the way, do you see this patch series (as whole or at least partially) in 4.8? Should I repost them before hard freeze? However, I am not sure that it (v7) can be taken into 4.8 because we are after last posting date. Wei, Jan, what is your standing in that case? Daniel
On Mon, Sep 19, 2016 at 04:27:11PM +0200, Daniel Kiper wrote: > On Mon, Sep 19, 2016 at 05:58:46AM -0600, Jan Beulich wrote: > > >>> On 12.09.16 at 22:18, <daniel.kiper@oracle.com> wrote: > > > --- a/xen/arch/x86/domain_page.c > > > +++ b/xen/arch/x86/domain_page.c > > > @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) > > > * domain's page tables but current may point at another domain's VCPU. > > > * Return NULL as though current is not properly set up yet. > > > */ > > > - if ( efi_enabled && efi_rs_using_pgtables() ) > > > + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) > > > > I think the efi_enabled() here is pointless now. > > Nope, it seems that Xen will blow up on BUG() in > xen/arch/x86/efi/stub.c:efi_rs_using_pgtables() if > compiler/linker cannot be used to build proper PE binary. > Of course we can change efi_rs_using_pgtables() to > return false in such case. > > > With this dropped (unless you know of a reason that it needs to stay) > > Reviewed-by: Jan Beulich <jbeulich@suse.com> > > Thanks a lot! > > By the way, do you see this patch series (as whole or > at least partially) in 4.8? Should I repost them before > hard freeze? However, I am not sure that it (v7) can be > taken into 4.8 because we are after last posting date. > > Wei, Jan, what is your standing in that case? > If this series ready of course it can be applied from my PoV. The last posting date is more things that are entirely new, which is not applicable to this series (v7 already). (Note I haven't checked all patches in this series) Wei. > Daniel
>>> On 19.09.16 at 16:27, <daniel.kiper@oracle.com> wrote: > On Mon, Sep 19, 2016 at 05:58:46AM -0600, Jan Beulich wrote: >> >>> On 12.09.16 at 22:18, <daniel.kiper@oracle.com> wrote: >> > --- a/xen/arch/x86/domain_page.c >> > +++ b/xen/arch/x86/domain_page.c >> > @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) >> > * domain's page tables but current may point at another domain's > VCPU. >> > * Return NULL as though current is not properly set up yet. >> > */ >> > - if ( efi_enabled && efi_rs_using_pgtables() ) >> > + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) >> >> I think the efi_enabled() here is pointless now. > > Nope, it seems that Xen will blow up on BUG() in > xen/arch/x86/efi/stub.c:efi_rs_using_pgtables() if > compiler/linker cannot be used to build proper PE binary. Ah, true. > Of course we can change efi_rs_using_pgtables() to > return false in such case. Except that it does already. You mean dropping the BUG() I guess. Both ways should be fine then for now. > By the way, do you see this patch series (as whole or > at least partially) in 4.8? Should I repost them before > hard freeze? However, I am not sure that it (v7) can be > taken into 4.8 because we are after last posting date. > > Wei, Jan, what is your standing in that case? We're not past that point afaik - we're past the last-posting-of- new-stuff date. Jan
On Mon, Sep 19, 2016 at 08:57:02AM -0600, Jan Beulich wrote: > >>> On 19.09.16 at 16:27, <daniel.kiper@oracle.com> wrote: > > On Mon, Sep 19, 2016 at 05:58:46AM -0600, Jan Beulich wrote: > >> >>> On 12.09.16 at 22:18, <daniel.kiper@oracle.com> wrote: > >> > --- a/xen/arch/x86/domain_page.c > >> > +++ b/xen/arch/x86/domain_page.c > >> > @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) > >> > * domain's page tables but current may point at another domain's > > VCPU. > >> > * Return NULL as though current is not properly set up yet. > >> > */ > >> > - if ( efi_enabled && efi_rs_using_pgtables() ) > >> > + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) > >> > >> I think the efi_enabled() here is pointless now. > > > > Nope, it seems that Xen will blow up on BUG() in > > xen/arch/x86/efi/stub.c:efi_rs_using_pgtables() if > > compiler/linker cannot be used to build proper PE binary. > > Ah, true. > > > Of course we can change efi_rs_using_pgtables() to > > return false in such case. > > Except that it does already. You mean dropping the BUG() I guess. Yep, right. However, should not we change "return 0" to "return false" in the same patch if efi_rs_using_pgtables() returns bool_t? > Both ways should be fine then for now. Dropping the BUG() from efi_rs_using_pgtables() and efi_enabled in above mentioned conditional makes final code, IMO, better. However, in this patch context it may look a bit strange. Is it acceptable for you to do both in this patch? > > By the way, do you see this patch series (as whole or > > at least partially) in 4.8? Should I repost them before > > hard freeze? However, I am not sure that it (v7) can be > > taken into 4.8 because we are after last posting date. > > > > Wei, Jan, what is your standing in that case? > > We're not past that point afaik - we're past the last-posting-of- > new-stuff date. Thanks for clarification guys. Daniel
>>> On 19.09.16 at 17:38, <daniel.kiper@oracle.com> wrote: > On Mon, Sep 19, 2016 at 08:57:02AM -0600, Jan Beulich wrote: >> >>> On 19.09.16 at 16:27, <daniel.kiper@oracle.com> wrote: >> > On Mon, Sep 19, 2016 at 05:58:46AM -0600, Jan Beulich wrote: >> >> >>> On 12.09.16 at 22:18, <daniel.kiper@oracle.com> wrote: >> >> > --- a/xen/arch/x86/domain_page.c >> >> > +++ b/xen/arch/x86/domain_page.c >> >> > @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) >> >> > * domain's page tables but current may point at another domain's >> > VCPU. >> >> > * Return NULL as though current is not properly set up yet. >> >> > */ >> >> > - if ( efi_enabled && efi_rs_using_pgtables() ) >> >> > + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) >> >> >> >> I think the efi_enabled() here is pointless now. >> > >> > Nope, it seems that Xen will blow up on BUG() in >> > xen/arch/x86/efi/stub.c:efi_rs_using_pgtables() if >> > compiler/linker cannot be used to build proper PE binary. >> >> Ah, true. >> >> > Of course we can change efi_rs_using_pgtables() to >> > return false in such case. >> >> Except that it does already. You mean dropping the BUG() I guess. > > Yep, right. However, should not we change "return 0" to "return false" > in the same patch if efi_rs_using_pgtables() returns bool_t? Since you don't touch that line anyway, and since using false there would then also call for switching from bool_t to bool, I'd rather leave this for another day. >> Both ways should be fine then for now. > > Dropping the BUG() from efi_rs_using_pgtables() and efi_enabled in > above mentioned conditional makes final code, IMO, better. However, > in this patch context it may look a bit strange. Is it acceptable > for you to do both in this patch? Yes, that what I tried to express with my previous reply. Jan
diff --git a/xen/arch/x86/dmi_scan.c b/xen/arch/x86/dmi_scan.c index b049e31..8dcb640 100644 --- a/xen/arch/x86/dmi_scan.c +++ b/xen/arch/x86/dmi_scan.c @@ -238,7 +238,7 @@ const char *__init dmi_get_table(paddr_t *base, u32 *len) { static unsigned int __initdata instance; - if (efi_enabled) { + if (efi_enabled(EFI_BOOT)) { if (efi_smbios3_size && !(instance & 1)) { *base = efi_smbios3_address; *len = efi_smbios3_size; @@ -696,7 +696,7 @@ static void __init dmi_decode(struct dmi_header *dm) void __init dmi_scan_machine(void) { - if ((!efi_enabled ? dmi_iterate(dmi_decode) : + if ((!efi_enabled(EFI_BOOT) ? dmi_iterate(dmi_decode) : dmi_efi_iterate(dmi_decode)) == 0) dmi_check_system(dmi_blacklist); else diff --git a/xen/arch/x86/domain_page.c b/xen/arch/x86/domain_page.c index d86f8fe..7541b91 100644 --- a/xen/arch/x86/domain_page.c +++ b/xen/arch/x86/domain_page.c @@ -36,7 +36,7 @@ static inline struct vcpu *mapcache_current_vcpu(void) * domain's page tables but current may point at another domain's VCPU. * Return NULL as though current is not properly set up yet. */ - if ( efi_enabled && efi_rs_using_pgtables() ) + if ( efi_enabled(EFI_RS) && efi_rs_using_pgtables() ) return NULL; /* diff --git a/xen/arch/x86/efi/stub.c b/xen/arch/x86/efi/stub.c index 07c2bd0..0bfaa74 100644 --- a/xen/arch/x86/efi/stub.c +++ b/xen/arch/x86/efi/stub.c @@ -4,9 +4,10 @@ #include <xen/lib.h> #include <asm/page.h> -#ifndef efi_enabled -const bool_t efi_enabled = 0; -#endif +bool efi_enabled(unsigned int feature) +{ + return false; +} void __init efi_init_memory(void) { } diff --git a/xen/arch/x86/mpparse.c b/xen/arch/x86/mpparse.c index ef6557c..c3d5bdc 100644 --- a/xen/arch/x86/mpparse.c +++ b/xen/arch/x86/mpparse.c @@ -564,7 +564,7 @@ static inline void __init construct_default_ISA_mptable(int mpc_default_type) static __init void efi_unmap_mpf(void) { - if (efi_enabled) + if (efi_enabled(EFI_BOOT)) clear_fixmap(FIX_EFI_MPF); } @@ -722,7 +722,7 @@ void __init find_smp_config (void) { unsigned int address; - if (efi_enabled) { + if (efi_enabled(EFI_BOOT)) { efi_check_config(); return; } diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c index 8ae897a..06f3970 100644 --- a/xen/arch/x86/setup.c +++ b/xen/arch/x86/setup.c @@ -439,8 +439,8 @@ static void __init parse_video_info(void) { struct boot_video_info *bvi = &bootsym(boot_vid_info); - /* The EFI loader fills vga_console_info directly. */ - if ( efi_enabled ) + /* vga_console_info is filled directly on EFI platform. */ + if ( efi_enabled(EFI_BOOT) ) return; if ( (bvi->orig_video_isVGA == 1) && (bvi->orig_video_mode == 3) ) @@ -726,7 +726,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) if ( !(mbi->flags & MBI_MODULES) || (mbi->mods_count == 0) ) panic("dom0 kernel not specified. Check bootloader configuration."); - if ( efi_enabled ) + if ( efi_enabled(EFI_LOADER) ) { set_pdx_range(xen_phys_start >> PAGE_SHIFT, (xen_phys_start + BOOTSTRAP_MAP_BASE) >> PAGE_SHIFT); @@ -741,6 +741,8 @@ void __init noreturn __start_xen(unsigned long mbi_p) memmap_type = loader; } + else if ( efi_enabled(EFI_BOOT) ) + memmap_type = "EFI"; else if ( e820_raw_nr != 0 ) { memmap_type = "Xen-e820"; @@ -837,7 +839,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) * we can relocate the dom0 kernel and other multiboot modules. Also, on * x86/64, we relocate Xen to higher memory. */ - for ( i = 0; !efi_enabled && i < mbi->mods_count; i++ ) + for ( i = 0; !efi_enabled(EFI_LOADER) && i < mbi->mods_count; i++ ) { if ( mod[i].mod_start & (PAGE_SIZE - 1) ) panic("Bootloader didn't honor module alignment request."); @@ -1078,7 +1080,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) if ( !xen_phys_start ) panic("Not enough memory to relocate Xen."); - reserve_e820_ram(&boot_e820, efi_enabled ? mbi->mem_upper : __pa(&_start), + reserve_e820_ram(&boot_e820, efi_enabled(EFI_LOADER) ? mbi->mem_upper : __pa(&_start), __pa(&_end)); /* Late kexec reservation (dynamic start address). */ diff --git a/xen/arch/x86/shutdown.c b/xen/arch/x86/shutdown.c index 0e1499d..54c2c79 100644 --- a/xen/arch/x86/shutdown.c +++ b/xen/arch/x86/shutdown.c @@ -116,7 +116,7 @@ void machine_halt(void) static void default_reboot_type(void) { if ( reboot_type == BOOT_INVALID ) - reboot_type = efi_enabled ? BOOT_EFI + reboot_type = efi_enabled(EFI_RS) ? BOOT_EFI : acpi_disabled ? BOOT_KBD : BOOT_ACPI; } diff --git a/xen/arch/x86/time.c b/xen/arch/x86/time.c index 73e0f98..f043998 100644 --- a/xen/arch/x86/time.c +++ b/xen/arch/x86/time.c @@ -727,7 +727,7 @@ static unsigned long get_cmos_time(void) static bool_t __read_mostly cmos_rtc_probe; boolean_param("cmos-rtc-probe", cmos_rtc_probe); - if ( efi_enabled ) + if ( efi_enabled(EFI_RS) ) { res = efi_get_time(); if ( res ) diff --git a/xen/common/efi/boot.c b/xen/common/efi/boot.c index 125c9ce..56544dc 100644 --- a/xen/common/efi/boot.c +++ b/xen/common/efi/boot.c @@ -934,6 +934,13 @@ efi_start(EFI_HANDLE ImageHandle, EFI_SYSTEM_TABLE *SystemTable) char *option_str; bool_t use_cfg_file; + __set_bit(EFI_BOOT, &efi_flags); + __set_bit(EFI_LOADER, &efi_flags); + +#ifndef CONFIG_ARM /* Disabled until runtime services implemented. */ + __set_bit(EFI_RS, &efi_flags); +#endif + efi_init(ImageHandle, SystemTable); use_cfg_file = efi_arch_use_config_file(SystemTable); @@ -1153,7 +1160,6 @@ efi_start(EFI_HANDLE ImageHandle, EFI_SYSTEM_TABLE *SystemTable) #ifndef CONFIG_ARM /* TODO - runtime service support */ -static bool_t __initdata efi_rs_enable = 1; static bool_t __initdata efi_map_uc; static void __init parse_efi_param(char *s) @@ -1171,7 +1177,12 @@ static void __init parse_efi_param(char *s) *ss = '\0'; if ( !strcmp(s, "rs") ) - efi_rs_enable = val; + { + if ( val ) + __set_bit(EFI_RS, &efi_flags); + else + __clear_bit(EFI_RS, &efi_flags); + } else if ( !strcmp(s, "attr=uc") ) efi_map_uc = val; @@ -1254,7 +1265,7 @@ void __init efi_init_memory(void) desc->PhysicalStart, desc->PhysicalStart + len - 1, desc->Type, desc->Attribute); - if ( !efi_rs_enable || + if ( !efi_enabled(EFI_RS) || (!(desc->Attribute & EFI_MEMORY_RUNTIME) && (!map_bs || (desc->Type != EfiBootServicesCode && @@ -1328,7 +1339,7 @@ void __init efi_init_memory(void) } } - if ( !efi_rs_enable ) + if ( !efi_enabled(EFI_RS) ) { efi_fw_vendor = NULL; return; diff --git a/xen/common/efi/runtime.c b/xen/common/efi/runtime.c index c256814..5f2de80 100644 --- a/xen/common/efi/runtime.c +++ b/xen/common/efi/runtime.c @@ -10,12 +10,6 @@ DEFINE_XEN_GUEST_HANDLE(CHAR16); #ifndef COMPAT -/* - * Currently runtime services are not implemented on ARM. To boot Xen with ACPI, - * set efi_enabled to 1, so that Xen can get the ACPI root pointer from EFI. - */ -const bool_t efi_enabled = 1; - #ifndef CONFIG_ARM # include <asm/i387.h> # include <asm/xstate.h> @@ -43,6 +37,9 @@ UINT64 __read_mostly efi_boot_max_var_store_size; UINT64 __read_mostly efi_boot_remain_var_store_size; UINT64 __read_mostly efi_boot_max_var_size; +/* Bit field representing available EFI features/properties. */ +unsigned int efi_flags; + struct efi __read_mostly efi = { .acpi = EFI_INVALID_TABLE_ADDR, .acpi20 = EFI_INVALID_TABLE_ADDR, @@ -53,6 +50,11 @@ struct efi __read_mostly efi = { const struct efi_pci_rom *__read_mostly efi_pci_roms; +bool efi_enabled(unsigned int feature) +{ + return test_bit(feature, &efi_flags); +} + #ifndef CONFIG_ARM /* TODO - disabled until implemented on ARM */ unsigned long efi_rs_enter(void) { diff --git a/xen/common/version.c b/xen/common/version.c index 0d31e38..223cb52 100644 --- a/xen/common/version.c +++ b/xen/common/version.c @@ -160,7 +160,7 @@ static int __init xen_build_init(void) #ifdef CONFIG_X86 /* Alternatively we may have a CodeView record from an EFI build. */ - if ( rc && efi_enabled ) + if ( rc && efi_enabled(EFI_LOADER) ) { const struct pe_external_debug_directory *dir = (const void *)n; diff --git a/xen/drivers/acpi/osl.c b/xen/drivers/acpi/osl.c index 9a49029..3616dfd 100644 --- a/xen/drivers/acpi/osl.c +++ b/xen/drivers/acpi/osl.c @@ -66,7 +66,7 @@ void __init acpi_os_vprintf(const char *fmt, va_list args) acpi_physical_address __init acpi_os_get_root_pointer(void) { - if (efi_enabled) { + if (efi_enabled(EFI_BOOT)) { if (efi.acpi20 != EFI_INVALID_TABLE_ADDR) return efi.acpi20; else if (efi.acpi != EFI_INVALID_TABLE_ADDR) diff --git a/xen/include/xen/efi.h b/xen/include/xen/efi.h index e74dad1..68c68a8 100644 --- a/xen/include/xen/efi.h +++ b/xen/include/xen/efi.h @@ -5,10 +5,13 @@ #include <xen/types.h> #endif -extern const bool_t efi_enabled; - #define EFI_INVALID_TABLE_ADDR (~0UL) +extern unsigned int efi_flags; +#define EFI_BOOT 0 /* Were we booted from EFI? */ +#define EFI_LOADER 1 /* Were we booted directly from EFI loader? */ +#define EFI_RS 2 /* Can we use runtime services? */ + /* Add fields here only if they need to be referenced from non-EFI code. */ struct efi { unsigned long mps; /* MPS table */ @@ -28,6 +31,7 @@ union compat_pf_efi_info; struct xenpf_efi_runtime_call; struct compat_pf_efi_runtime_call; +bool efi_enabled(unsigned int feature); void efi_init_memory(void); bool_t efi_rs_using_pgtables(void); unsigned long efi_get_time(void);
First of all we need to differentiate between legacy BIOS and EFI platforms during runtime, not during build, because one image will have legacy and EFI code and can be executed on both platforms. Additionally, we need more fine grained knowledge about EFI environment and check for EFI platform and EFI loader separately to properly support multiboot2 protocol. In general Xen loaded by this protocol uses memory mappings and loaded modules in similar way to Xen loaded by multiboot (v1) protocol. Hence, create efi_enabled() which checks available features in efi_flags. This patch defines EFI_BOOT, EFI_LOADER and EFI_RS features. EFI_BOOT is equal to old efi_enabled == 1. EFI_RS ease control on runtime services usage. EFI_LOADER tells that Xen was loaded directly from EFI as PE executable. Suggested-by: Jan Beulich <jbeulich@suse.com> Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com> --- v6 - suggestions/fixes: - define efi_enabled() as "bool efi_enabled(unsigned int feature)" instead of "bool_t efi_enabled(int feature)" (suggested by Jan Beulich), - define efi_flags as unsigned int (suggested by Jan Beulich), - various minor cleanups and fixes (suggested by Jan Beulich). v5 - suggestions/fixes: - squash three patches into one (suggested by Jan Beulich), - introduce all features at once (suggested by Jan Beulich), - efi_enabled() returns bool_t instead of unsigned int (suggested by Jan Beulich), - update commit message. v4 - suggestions/fixes: - rename EFI_PLATFORM to EFI_BOOT (suggested by Jan Beulich), - move EFI_BOOT definition to efi struct definition (suggested by Jan Beulich), - remove unneeded efi.flags initialization (suggested by Jan Beulich), - use __set_bit() instead of set_bit() if possible (suggested by Jan Beulich), - do efi_enabled() cleanup (suggested by Jan Beulich), - improve comments (suggested by Jan Beulich), - improve commit message. v3 - suggestions/fixes: - define efi struct in xen/arch/x86/efi/stub.c in earlier patch (suggested by Jan Beulich), - improve comments (suggested by Jan Beulich), - improve commit message (suggested by Jan Beulich). --- xen/arch/x86/dmi_scan.c | 4 ++-- xen/arch/x86/domain_page.c | 2 +- xen/arch/x86/efi/stub.c | 7 ++++--- xen/arch/x86/mpparse.c | 4 ++-- xen/arch/x86/setup.c | 12 +++++++----- xen/arch/x86/shutdown.c | 2 +- xen/arch/x86/time.c | 2 +- xen/common/efi/boot.c | 19 +++++++++++++++---- xen/common/efi/runtime.c | 14 ++++++++------ xen/common/version.c | 2 +- xen/drivers/acpi/osl.c | 2 +- xen/include/xen/efi.h | 8 ++++++-- 12 files changed, 49 insertions(+), 29 deletions(-)