| Message ID | 20210910021236.15082-2-dpsmith@apertussolutions.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | xsm: refactoring xsm hooks | expand |
On 10.09.2021 04:12, Daniel P. Smith wrote: > From: Andrew Cooper <andrew.cooper3@citrix.com> > > The alternative call infrastructure is x86-only for now, but the common iommu > code has a variant and more common code wants to use the infrastructure. > > Introduce CONFIG_ALTERNATIVE_CALL and a conditional implemetnation so common > code can use the optimisation when available, without requiring all > architectures to implement no-op stubs. > > Write some documentation, which was thus far entirely absent, covering the > requirements for an architecture to implement this optimsiation, and how to > use the infrastructure in general code. > > Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> > Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com> I did offer an A-b on v4, dependent upon two small adjustments. Was it deliberate to neither make these adjustments nor add the ack? (At that point I did offer making the adjustments while committing, but I think it goes without saying that upon re-submission the changes should be made. Unless of course there's disagreement about making them in the first place. Yet such disagreement would be worth making explicit then, imo.) Jan
On 9/10/21 2:10 AM, Jan Beulich wrote: > On 10.09.2021 04:12, Daniel P. Smith wrote: >> From: Andrew Cooper <andrew.cooper3@citrix.com> >> >> The alternative call infrastructure is x86-only for now, but the common iommu >> code has a variant and more common code wants to use the infrastructure. >> >> Introduce CONFIG_ALTERNATIVE_CALL and a conditional implemetnation so common >> code can use the optimisation when available, without requiring all >> architectures to implement no-op stubs. >> >> Write some documentation, which was thus far entirely absent, covering the >> requirements for an architecture to implement this optimsiation, and how to >> use the infrastructure in general code. >> >> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> >> Signed-off-by: Daniel P. Smith <dpsmith@apertussolutions.com> > > I did offer an A-b on v4, dependent upon two small adjustments. Was it > deliberate to neither make these adjustments nor add the ack? (At that > point I did offer making the adjustments while committing, but I think > it goes without saying that upon re-submission the changes should be > made. Unless of course there's disagreement about making them in the > first place. Yet such disagreement would be worth making explicit then, > imo.) > > Jan > Yes you did, perhaps I did not follow the thread correctly. Re-reading it you mentioned they could be addressed during committing. Since I am going to have to respin this because the hard-tab foul up, I can pick up those to adjustments and add your Ack-by. v/r, dps
diff --git a/xen/arch/x86/Kconfig b/xen/arch/x86/Kconfig index 9b164db641..1f83518ee0 100644 --- a/xen/arch/x86/Kconfig +++ b/xen/arch/x86/Kconfig @@ -6,6 +6,7 @@ config X86 def_bool y select ACPI select ACPI_LEGACY_TABLES_LOOKUP + select ALTERNATIVE_CALL select ARCH_SUPPORTS_INT128 select CORE_PARKING select HAS_ALTERNATIVE diff --git a/xen/common/Kconfig b/xen/common/Kconfig index 0ddd18e11a..ac5491b1cc 100644 --- a/xen/common/Kconfig +++ b/xen/common/Kconfig @@ -22,6 +22,9 @@ config GRANT_TABLE If unsure, say Y. +config ALTERNATIVE_CALL + bool + config HAS_ALTERNATIVE bool diff --git a/xen/include/xen/alternative-call.h b/xen/include/xen/alternative-call.h new file mode 100644 index 0000000000..d10af90b1b --- /dev/null +++ b/xen/include/xen/alternative-call.h @@ -0,0 +1,63 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef XEN_ALTERNATIVE_CALL +#define XEN_ALTERNATIVE_CALL + +/* + * Some subsystems in Xen may have multiple implementions, which can be + * resolved to a single implementation at boot time. By default, this will + * result in the use of function pointers. + * + * Some architectures may have mechanisms for dynamically modifying .text. + * Using this mechnaism, function pointers can be converted to direct calls + * which are typically more efficient at runtime. + * + * For architectures to support: + * + * - Implement alternative_{,v}call() in asm/alternative.h. Code generation + * requirements are to emit a function pointer call at build time, and stash + * enough metadata to simplify the call at boot once the implementation has + * been resolved. + * - Select ALTERNATIVE_CALL in Kconfig. + * + * To use: + * + * Consider the following simplified example. + * + * 1) struct foo_ops __alt_call_maybe_initdata ops; + * + * 2) const struct foo_ops __initconst foo_a_ops = { ... }; + * const struct foo_ops __initconst foo_b_ops = { ... }; + * + * void foo_init(void) + * { + * ... + * if ( use_impl_a ) + * ops = *foo_a_ops; + * else if ( use_impl_b ) + * ops = *foo_b_ops; + * ... + * } + * + * 3) alternative_call(ops.bar, ...); + * + * There needs to a single ops object (1) which will eventually contain the + * function pointers. This should be populated in foo's init() function (2) + * by one of the available implementations. To call functions, use + * alternative_{,v}call() referencing the main ops object (3). + */ + +#ifdef CONFIG_ALTERNATIVE_CALL + +#include <asm/alternative.h> + +#define __alt_call_maybe_initdata __initdata + +#else + +#define alternative_call(func, args...) (func)(args) +#define alternative_vcall(func, args...) (func)(args) + +#define __alt_call_maybe_initdata __read_mostly + +#endif /* !CONFIG_ALTERNATIVE_CALL */ +#endif /* XEN_ALTERNATIVE_CALL */