| Message ID | 20240717111817.3517937-1-andrew.cooper3@citrix.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [for-4.20] hvmloader: Use fastcall everywhere | expand |
On 17.07.2024 13:18, Andrew Cooper wrote: > HVMLoader is a single freestanding 32bit program with no external > dependencies. Use the fastcall calling convetion (up to 3 parameters in > registers rather than on the stack) globally, which is more efficient because > of reducing the number of stack operations. > > Some bloat-o-meter highligs are: > > add/remove: 0/0 grow/shrink: 3/118 up/down: 8/-3004 (-2996) > Function old new delta > ... > hvmloader_acpi_build_tables 1125 961 -164 > acpi_build_tables 1277 1081 -196 > pci_setup 4756 4516 -240 > construct_secondary_tables 1689 1447 -242 > > No functional change. > > Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> Reviewed-by: Jan Beulich <jbeulich@suse.com>
diff --git a/tools/firmware/hvmloader/Makefile b/tools/firmware/hvmloader/Makefile index 0e1dce26b342..05572f17f609 100644 --- a/tools/firmware/hvmloader/Makefile +++ b/tools/firmware/hvmloader/Makefile @@ -25,7 +25,7 @@ ld-option = $(shell if $(LD) -v $(1) >/dev/null 2>&1; then echo y; else echo n; # SMBIOS spec requires format mm/dd/yyyy SMBIOS_REL_DATE ?= $(shell date +%m/%d/%Y) -CFLAGS += $(CFLAGS_xeninclude) -fno-pic +CFLAGS += $(CFLAGS_xeninclude) -fno-pic -mregparm=3 # We mustn't use tools-only public interfaces. CFLAGS += -D__XEN_INTERFACE_VERSION__=__XEN_LATEST_INTERFACE_VERSION__ diff --git a/tools/firmware/hvmloader/smp.c b/tools/firmware/hvmloader/smp.c index 5d46eee1c5f4..1b940cefd071 100644 --- a/tools/firmware/hvmloader/smp.c +++ b/tools/firmware/hvmloader/smp.c @@ -31,7 +31,7 @@ static int ap_callin; -static void __attribute__((regparm(1))) cpu_setup(unsigned int cpu) +static void cpu_setup(unsigned int cpu) { printf(" - CPU%d ... ", cpu); cacheattr_init();
HVMLoader is a single freestanding 32bit program with no external dependencies. Use the fastcall calling convetion (up to 3 parameters in registers rather than on the stack) globally, which is more efficient because of reducing the number of stack operations. Some bloat-o-meter highligs are: add/remove: 0/0 grow/shrink: 3/118 up/down: 8/-3004 (-2996) Function old new delta ... hvmloader_acpi_build_tables 1125 961 -164 acpi_build_tables 1277 1081 -196 pci_setup 4756 4516 -240 construct_secondary_tables 1689 1447 -242 No functional change. Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> --- CC: Jan Beulich <JBeulich@suse.com> CC: Roger Pau Monné <roger.pau@citrix.com> I was surprised to find that nowhere in the handover to other blobs do we pass parameters on the stack, but then again the entry point of the new blobs typically happen with no stack anyway so I suppose I shouldn't be surprised. --- tools/firmware/hvmloader/Makefile | 2 +- tools/firmware/hvmloader/smp.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) base-commit: 57338346f29cea7b183403561bdc5f407163b846 prerequisite-patch-id: 212002cd69805ec19f556756d487638c7d348878