riscv: fix --enable-debug in riscv-to-apply.next
cover riscv: fix --enable-debug in riscv-to-apply.next
Hi,
There is a couple of build problems that sneaked in riscv-to-apply.next in the KVM AIA patches. Here's a potential solution for them.
Alistair, feel free to squash the fixes in the original commits since they're not yet on master.
Daniel Henrique Barboza (2):
hw/intc/riscv_aplic.c fix non-KVM --enable-debug build hw/riscv/virt.c: fix non-KVM --enable-debug build
hw/intc/riscv_aplic.c | 2 ++ target/riscv/kvm_riscv.h | 10 ++++++++++ 2 files changed, 12 insertions(+)
Signature · 1 line
2.41.0
PATCH 1/2 hw/intc/riscv_aplic.c fix non-KVM --enable-debug build
Commit 6df0b37e2ab breaks a --enable-debug build in a non-KVM environment with the following error:
/usr/bin/ld: libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':./qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'
collect2: error: ld returned 1 exit status
This happens because the debug build will poke into the
'if (is_kvm_aia(aplic->msimode))' block and fail to find a reference to the KVM only function riscv_kvm_aplic_request().
Make riscv_kvm_aplic_request() a no-op if we're not building KVM.
--- hw/intc/riscv_aplic.c | 2 ++ 1 file changed, 2 insertions(+)
hw/intc/riscv_aplic.c +2 -0
@@ -483,7 +483,9 @@ static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc)static void riscv_kvm_aplic_request (void *opaque,int irq,int level) { +#ifdef CONFIG_KVM kvm_set_irq (kvm_state, irq, !!level); +#endif }static void riscv_aplic_request (void *opaque,int irq,int level)
Signature · 1 line
2.41.0
Re: [PATCH 1/2] hw/intc/riscv_aplic.c fix non-KVM --enable-debug build
Quoted 24 lines · Daniel Henrique Barboza
Commit
6df0b37e2abbreaks a --enable-debug build in a non-KVM environment with the following error:
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':./qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'
collect2: error: ld returned 1 exit statusThis happens because the debug build will poke into the
'if (is_kvm_aia(aplic->msimode))' block and fail to find a reference to the KVM only functionriscv_kvm_aplic_request().Make
riscv_kvm_aplic_request()a no-op if we're not building KVM.Fixes 6df0b37e2ab ("target/riscv: update APLIC and IMSIC to support KVM AIA")Signed-off-by Daniel Henrique Barboza <[email protected]>--- hw/intc/riscv_aplic.c | 2 ++ 1 file changed, 2 insertions(+)hw/intc/riscv_aplic.c +0
-0
hw/intc/riscv_aplic.c static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc) · line 489
Quoted 7 lines
hw/intc/riscv_aplic.c +2
-0@@ -483,7 +483,9 @@ static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc)static void riscv_kvm_aplic_request (void *opaque,int irq,int level) { +#ifdef CONFIG_KVM kvm_set_irq (kvm_state, irq, !!level); +#endif }
Can we use kvm_enabled() to let the compiler elide KVM code when !CONFIG_KVM?
-- >8 --
hw/intc/riscv_aplic.c +1 -2
@@ -155,7 +155,7 @@*/ static bool is_kvm_aia (bool msimode) { -return kvm_irqchip_in_kernel () && msimode; +return kvm_enabled () &&kvm_irqchip_in_kernel () && msimode; } ---
Re: [PATCH 1/2] hw/intc/riscv_aplic.c fix non-KVM --enable-debug build
Quoted 39 lines · Philippe Mathieu-Daudé
Quoted 24 lines · Daniel Henrique Barboza
Commit
6df0b37e2abbreaks a --enable-debug build in a non-KVM environment with the following error:
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':./qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'
collect2: error: ld returned 1 exit statusThis happens because the debug build will poke into the
'if (is_kvm_aia(aplic->msimode))' block and fail to find a reference to the KVM only functionriscv_kvm_aplic_request().Make
riscv_kvm_aplic_request()a no-op if we're not building KVM.Fixes 6df0b37e2ab ("target/riscv: update APLIC and IMSIC to support KVM AIA")Signed-off-by Daniel Henrique Barboza <[email protected]>---
hw/intc/riscv_aplic.c| 2 ++1 file changed, 2 insertions(+)hw/intc/riscv_aplic.c +0
-0hw/intc/riscv_aplic.c static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc) · line 488
Quoted 6 lines
hw/intc/riscv_aplic.c +2
-0@@ -483,7 +483,9 @@ static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc) static void riscv_kvm_aplic_request(void *opaque, int irq, int level) { +#ifdef CONFIG_KVM kvm_set_irq(kvm_state, irq, !!level); +#endif }Can we use
kvm_enabled()to let the compiler elide KVM code when !CONFIG_KVM?-- >8 --
hw/intc/riscv_aplic.c +0
-0
hw/intc/riscv_aplic.c line 158
Quoted 5 lines
hw/intc/riscv_aplic.c +1
-1@@ -155,7 +155,7 @@*/ static bool is_kvm_aia(bool msimode) { - return kvm_irqchip_in_kernel() && msimode; + return kvm_enabled() && kvm_irqchip_in_kernel() && msimode;
Apparently we can't:
$ git diff
hw/intc/riscv_aplic.c +1 -3
@@ -155,7 +155,7 @@*/ static bool is_kvm_aia (bool msimode) { -return kvm_irqchip_in_kernel () && msimode; +return kvm_enabled () &&kvm_irqchip_in_kernel () && msimode; } static uint32_t riscv_aplic_read_input_word(RISCVAPLICState *aplic,
@@ -483,9 +483,7 @@ static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc)static void riscv_kvm_aplic_request (void *opaque,int irq,int level) { -#ifdef CONFIG_KVM kvm_set_irq (kvm_state, irq, !!level); -#endif }
Same error appears:
/usr/bin/ld: libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':/home/danielhb/work/qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'
collect2: error: ld returned 1 exit status
Thanks,
Daniel
Quoted 3 lines
hw/intc/riscv_aplic.c +0
-1} ---
PATCH 2/2 hw/riscv/virt.c: fix non-KVM --enable-debug build
In reply to Daniel Henrique BarbozaA build with --enable-debug and without KVM will fail as follows:
/usr/bin/ld: libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':./qemu/build/../hw/riscv/virt.c:1465: undefined reference to `kvm_riscv_aia_create'
This happens because the code block with "if virt_use_kvm_aia(s)" isn't being ignored by the debug build, resulting in an undefined reference to a KVM only function.
Add a stub for kvm_riscv_aia_create() in kvm_riscv.h when CONFIG_KVM is false. Adding it as an inline instead of using kvm-stubs.c will make it easier in the future to remember to add stubs for kvm functions that are used in multiple accelerator code.
--- target/riscv/kvm_riscv.h | 10 ++++++++++ 1 file changed, 10 insertions(+)
target/riscv/kvm_riscv.h +10 -0
@@ -22,9 +22,19 @@void kvm_riscv_init_user_properties (Object *cpu_obj);void kvm_riscv_reset_vcpu (RISCVCPU *cpu);void kvm_riscv_set_irq (RISCVCPU *cpu,int irq,int level); + +#ifdef CONFIG_KVM void kvm_riscv_aia_create (MachineState *machine,uint64_t group_shift,uint64_t aia_irq_num,uint64_t aia_msi_num,uint64_t aplic_base,uint64_t imsic_base,uint64_t guest_num); +#else +static inline void kvm_riscv_aia_create (MachineState *machine, +uint64_t group_shift,uint64_t aia_irq_num, +uint64_t aia_msi_num,uint64_t aplic_base, +uint64_t imsic_base,uint64_t guest_num) { +g_assert_not_reached (); +} +#endif #endif
Signature · 1 line
2.41.0
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 14 lines · Daniel Henrique Barboza
A build with --enable-debug and without KVM will fail as follows:
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':./qemu/build/../hw/riscv/virt.c:1465: undefined reference to `kvm_riscv_aia_create'This happens because the code block with "if
virt_use_kvm_aia(s)" isn't being ignored by the debug build, resulting in an undefined reference to a KVM only function.Add a stub for
kvm_riscv_aia_create()inkvm_riscv.hwhenCONFIG_KVMis false. Adding it as an inline instead of usingkvm-stubs.cwill make it easier in the future to remember to add stubs for kvm functions that are used in multiple accelerator code.
Quoted 5 lines
Fixes dbdb99948e ("target/riscv: select KVM AIA in riscv virt machine")Signed-off-by Daniel Henrique Barboza <[email protected]>--- target/riscv/kvm_riscv.h | 10 ++++++++++ 1 file changed, 10 insertions(+)
Similarly:
-- >8 --
hw/riscv/virt.c +0 -0
@@ -79,7 +79,9 @@/* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by
QEMU. */
static bool virt_use_kvm_aia(RISCVVirtState *s)
{
- return kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC;
+ return kvm_enabled()
+ && kvm_irqchip_in_kernel()
+ && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC;
}
---
?
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 26 lines · Philippe Mathieu-Daudé
Quoted 14 lines · Daniel Henrique Barboza
A build with --enable-debug and without KVM will fail as follows:
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':./qemu/build/../hw/riscv/virt.c:1465: undefined reference to `kvm_riscv_aia_create'This happens because the code block with "if
virt_use_kvm_aia(s)" isn't being ignored by the debug build, resulting in an undefined reference to a KVM only function.Add a stub for
kvm_riscv_aia_create()inkvm_riscv.hwhenCONFIG_KVMis false. Adding it as an inline instead of usingkvm-stubs.cwill make it easier in the future to remember to add stubs for kvm functions that are used in multiple accelerator code.Quoted 5 lines
Fixes dbdb99948e ("target/riscv: select KVM AIA in riscv virt machine")Signed-off-by Daniel Henrique Barboza <[email protected]>---
target/riscv/kvm_riscv.h| 10 ++++++++++1 file changed, 10 insertions(+)Similarly:
-- >8 --
hw/riscv/virt.c line 84
Quoted 9 lines
hw/riscv/virt.c +3
-2@@ -79,7 +79,9 @@ /* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by QEMU. */ static bool virt_use_kvm_aia(RISCVVirtState *s) { - return kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; + return kvm_enabled() + && kvm_irqchip_in_kernel() + && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; } ---
It doesn't work. Same error:
hw/riscv/virt.c +2 -1
@@ -79,7 +79,8 @@/* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by QEMU. */ static bool virt_use_kvm_aia (RISCVVirtState *s) { -return kvm_irqchip_in_kernel () && s->aia_type ==VIRT_AIA_TYPE_APLIC_IMSIC ; +return kvm_enabled () && +kvm_irqchip_in_kernel () && s->aia_type ==VIRT_AIA_TYPE_APLIC_IMSIC ; }static const MemMapEntry virt_memmap[] = {
target/riscv/kvm_riscv.h +0 -10
@@ -22,19 +22,9 @@void kvm_riscv_init_user_properties (Object *cpu_obj);void kvm_riscv_reset_vcpu (RISCVCPU *cpu);void kvm_riscv_set_irq (RISCVCPU *cpu,int irq,int level); - -#ifdef CONFIG_KVM void kvm_riscv_aia_create (MachineState *machine,uint64_t group_shift,uint64_t aia_irq_num,uint64_t aia_msi_num,uint64_t aplic_base,uint64_t imsic_base,uint64_t guest_num); -#else -static inline void kvm_riscv_aia_create (MachineState *machine, -uint64_t group_shift,uint64_t aia_irq_num, -uint64_t aia_msi_num,uint64_t aplic_base, -uint64_t imsic_base,uint64_t guest_num) { -g_assert_not_reached (); -} -#endif
/usr/bin/ld: libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':/home/danielhb/work/qemu/build/../hw/riscv/virt.c:1466: undefined reference to `kvm_riscv_aia_create'
collect2: error: ld returned 1 exit status
I'm no compiler expert by any means but it seems that the --enable-debug build does not strip things out like the usual build does, e.g. it won't elide a 'if kvm_enabled()' block out by checking that kvm_enabled() is always false.
Thanks,
Daniel
Quoted 2 lines
?
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 54 lines · Daniel Henrique Barboza
hw/riscv/virt.c line 84
Quoted 12 lines
-- >8 --
hw/riscv/virt.c +3
-2@@ -79,7 +79,9 @@ /* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by QEMU. */ static bool virt_use_kvm_aia(RISCVVirtState *s) { - return kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; + return kvm_enabled() + && kvm_irqchip_in_kernel() + && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; } ---It doesn't work. Same error:
hw/riscv/virt.c +2
-1@@ -79,7 +79,8 @@ /* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by QEMU. */ static bool virt_use_kvm_aia(RISCVVirtState *s) { - return kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; + return kvm_enabled() && + kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; } static const MemMapEntry virt_memmap[] = {target/riscv/kvm_riscv.h +0
-10@@ -22,19 +22,9 @@ void kvm_riscv_init_user_properties(Object *cpu_obj); void kvm_riscv_reset_vcpu(RISCVCPU *cpu); void kvm_riscv_set_irq(RISCVCPU *cpu, int irq, int level); - -#ifdef CONFIG_KVM void kvm_riscv_aia_create(MachineState *machine, uint64_t group_shift, uint64_t aia_irq_num, uint64_t aia_msi_num, uint64_t aplic_base, uint64_t imsic_base, uint64_t guest_num); -#else -static inline void kvm_riscv_aia_create (MachineState *machine, - uint64_t group_shift, uint64_t aia_irq_num, - uint64_t aia_msi_num, uint64_t aplic_base, - uint64_t imsic_base, uint64_t guest_num) { - g_assert_not_reached(); -} -#endif
Quoted 8 lines
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':/home/danielhb/work/qemu/build/../hw/riscv/virt.c:1466: undefined reference to `kvm_riscv_aia_create'
collect2: error: ld returned 1 exit statusI'm no compiler expert by any means but it seems that the --enable-debug build does not strip things out like the usual build does, e.g. it won't elide a 'if
kvm_enabled()' block out by checking thatkvm_enabled()is always false.
The compiler certainly does eliminate 0 && foo(), even at -O0.
There must be something else going on.
Pointer to your tree?
r~
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 60 lines · Richard Henderson
Quoted 59 lines · Daniel Henrique Barboza
hw/riscv/virt.c line 84
Quoted 12 lines
-- >8 --
hw/riscv/virt.c +3
-2@@ -79,7 +79,9 @@ /* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by QEMU. */ static bool virt_use_kvm_aia(RISCVVirtState *s) { - return kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; + return kvm_enabled() + && kvm_irqchip_in_kernel() + && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; } ---It doesn't work. Same error:
hw/riscv/virt.c +2
-1@@ -79,7 +79,8 @@ /* KVM AIA only supports APLIC MSI. APLIC Wired is always emulated by QEMU. */ static bool virt_use_kvm_aia(RISCVVirtState *s) { - return kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; + return kvm_enabled() && + kvm_irqchip_in_kernel() && s->aia_type == VIRT_AIA_TYPE_APLIC_IMSIC; } static const MemMapEntry virt_memmap[] = {target/riscv/kvm_riscv.h +0
-10@@ -22,19 +22,9 @@ void kvm_riscv_init_user_properties(Object *cpu_obj); void kvm_riscv_reset_vcpu(RISCVCPU *cpu); void kvm_riscv_set_irq(RISCVCPU *cpu, int irq, int level); - -#ifdef CONFIG_KVM void kvm_riscv_aia_create(MachineState *machine, uint64_t group_shift, uint64_t aia_irq_num, uint64_t aia_msi_num, uint64_t aplic_base, uint64_t imsic_base, uint64_t guest_num); -#else -static inline void kvm_riscv_aia_create (MachineState *machine, - uint64_t group_shift, uint64_t aia_irq_num, - uint64_t aia_msi_num, uint64_t aplic_base, - uint64_t imsic_base, uint64_t guest_num) { - g_assert_not_reached(); -} -#endif
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':/home/danielhb/work/qemu/build/../hw/riscv/virt.c:1466: undefined reference to `kvm_riscv_aia_create'
collect2: error: ld returned 1 exit status
Quoted 8 lines
Quoted 3 lines
I'm no compiler expert by any means but it seems that the --enable-debug build does not strip things out like the usual build does, e.g. it won't elide a 'if
kvm_enabled()' block out by checking thatkvm_enabled()is always false.The compiler certainly does eliminate 0 &&
foo(), even at -O0.There must be something else going on.
Pointer to your tree?
It's this tree:
https://github.com/alistair23/qemu/tree/riscv-to-apply.next
Building using --enable-debug fails:
[danielhb@grind build]$ ../configure --target-list=riscv64-softmmu --enable-debug && make -j
(...)
0.so /usr/lib64/libcairo-gobject.so /usr/lib64/libcairo.so /usr/lib64/libgdk_pixbuf-2.0.so /usr/lib64/libX11.so /usr/lib64/libvirglrenderer.so /usr/lib64/libcacard.so /usr/lib64/libusbredirparser.so /usr/lib64/libusb-1.0.so -lbrlapi @block.syms -lnuma /usr/lib64/liburing.so -lm /usr/lib64/libfuse3.so -lpthread /usr/lib64/iscsi/libiscsi.so -laio /usr/lib64/libcurl.so /usr/lib64/libnfs.so /usr/lib64/libssh.so -lrbd -lrados -lbz2 -lutil -Wl,--end-group /usr/bin/ld: libqemu-riscv64-softmmu.fa.p/hw_riscv_virt.c.o: in function `virt_machine_init':/home/danielhb/work/qemu/build/../hw/riscv/virt.c:1465: undefined reference to `kvm_riscv_aia_create'/usr/bin/ld: libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':/home/danielhb/work/qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'
collect2: error: ld returned 1 exit status
Building without --enable-debug works. gitlab CI doesn't seem to care with it because it's all green with this tree.
The commits that introduced these errors are listed in the 'Fixes' tag of patches 1 and 2.
Thanks,
Daniel
Quoted 2 lines
r~
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 8 lines · Daniel Henrique Barboza
Quoted 4 lines
The compiler certainly does eliminate 0 &&
foo(), even at -O0.There must be something else going on.
Pointer to your tree?It's this tree:
Ok, so while -O0 will eliminate 0 && foo(), it doesn't eliminate with bar() && foo(), where bar must be inlined (multiple times in this case) to find the 0.
Moreover in the case of
Quoted 2 lines
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':/home/danielhb/work/qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'
this one, where foo (aka riscv_kvm_aplic_request) would have to be eliminated as well. But the compiler won't eliminate entire unused functions with -O0.
This seems to do the trick. Whether it is aesthetically better than what you had with your patches, I will leave to someone else.
r~
hw/intc/riscv_aplic.c +4 -0
@@ -481,10 +481,14 @@ static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc)return topi; } +#ifdef CONFIG_KVM static void riscv_kvm_aplic_request (void *opaque,int irq,int level) {kvm_set_irq (kvm_state, irq, !!level); } +#else +#define riscv_kvm_aplic_request ({ qemu_build_not_reached(); NULL; }) +#endif static void riscv_aplic_request (void *opaque,int irq,int level) {
hw/riscv/virt.c +2 -2
@@ -782,7 +782,7 @@ static void create_fdt_sockets(RISCVVirtState *s, const MemMapEntry *memmap, }/* KVM AIA only has one APLIC instance */ -if (virt_use_kvm_aia (s)) { +if (kvm_enabled () &&virt_use_kvm_aia (s)) { create_fdt_socket_aplic(s, memmap, 0, msi_m_phandle, msi_s_phandle, phandle, &intc_phandles[0], xplic_phandles,
@@ -1461,7 +1461,7 @@ static void virt_machine_init(MachineState *machine) } } -if (virt_use_kvm_aia (s)) { +if (kvm_enabled () &&virt_use_kvm_aia (s)) { kvm_riscv_aia_create(machine, IMSIC_MMIO_GROUP_MIN_SHIFT, VIRT_IRQCHIP_NUM_SOURCES, VIRT_IRQCHIP_NUM_MSIS, memmap[VIRT_APLIC_S].base,
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 60 lines · Richard Henderson
Quoted 8 lines · Daniel Henrique Barboza
Quoted 4 lines
The compiler certainly does eliminate 0 &&
foo(), even at -O0.There must be something else going on.
Pointer to your tree?It's this tree:
Ok, so while -O0 will eliminate 0 &&
foo(), it doesn't eliminate withbar()&&foo(), where bar must be inlined (multiple times in this case) to find the 0.Moreover in the case of
Quoted 2 lines
/usr/bin/ld:libqemu-riscv64-softmmu.fa.p/hw_intc_riscv_aplic.c.o: in function `riscv_kvm_aplic_request':/home/danielhb/work/qemu/build/../hw/intc/riscv_aplic.c:486: undefined reference to `kvm_set_irq'this one, where foo (aka
riscv_kvm_aplic_request) would have to be eliminated as well. But the compiler won't eliminate entire unused functions with -O0.This seems to do the trick. Whether it is aesthetically better than what you had with your patches, I will leave to someone else.
r~
hw/intc/riscv_aplic.c +4
-0@@ -481,10 +481,14 @@ static uint32_t riscv_aplic_idc_claimi(RISCVAPLICState *aplic, uint32_t idc) return topi; } +#ifdef CONFIG_KVM static void riscv_kvm_aplic_request(void *opaque, int irq, int level) { kvm_set_irq(kvm_state, irq, !!level); } +#else +#define riscv_kvm_aplic_request ({ qemu_build_not_reached(); NULL; }) +#endif static void riscv_aplic_request(void *opaque, int irq, int level) {hw/riscv/virt.c +1
-1@@ -782,7 +782,7 @@ static void create_fdt_sockets(RISCVVirtState *s, const MemMapEntry *memmap, } /* KVM AIA only has one APLIC instance */ - if (virt_use_kvm_aia(s)) { + if (kvm_enabled() && virt_use_kvm_aia(s)) { create_fdt_socket_aplic(s, memmap, 0, msi_m_phandle, msi_s_phandle, phandle, &intc_phandles[0], xplic_phandles,
hw/riscv/virt.c static void virt_machine_init(MachineState *machine) · line 1467
Quoted 8 lines
hw/riscv/virt.c +1
-1@@ -1461,7 +1461,7 @@ static void virt_machine_init(MachineState *machine) } } - if (virt_use_kvm_aia(s)) { + if (kvm_enabled() && virt_use_kvm_aia(s)) { kvm_riscv_aia_create(machine, IMSIC_MMIO_GROUP_MIN_SHIFT, VIRT_IRQCHIP_NUM_SOURCES, VIRT_IRQCHIP_NUM_MSIS, memmap[VIRT_APLIC_S].base,
I'll leave to Alistair to decide, both seems good to me.
TBH I'm bothered why this doesn't work:
hw/intc/riscv_aplic.c +9 -5
@@ -839,12 +839,16 @@ static void riscv_aplic_realize(DeviceState *dev, Error **errp)* Only root APLICs have hardware IRQ lines. All non-root APLICs * have IRQ lines delegated by their parent APLIC. */ -if (!aplic->parent) { -if (is_kvm_aia (aplic->msimode)) { -qdev_init_gpio_in (dev, riscv_kvm_aplic_request, aplic->num_irqs); - }else { -qdev_init_gpio_in (dev, riscv_aplic_request, aplic->num_irqs); +if (kvm_enabled ()) { +if (!aplic->parent) { +if (is_kvm_aia (aplic->msimode)) { +qdev_init_gpio_in (dev, riscv_kvm_aplic_request, aplic->num_irqs); + }else { +qdev_init_gpio_in (dev, riscv_aplic_request, aplic->num_irqs); + } } + }else if (!aplic->parent) { +qdev_init_gpio_in (dev, riscv_aplic_request, aplic->num_irqs); }
Why is the compiler refusing to crop an "if kvm_enabled()" block? There's no other conditionals to handle, and it is able to crop "if (kvm_enabled() && virt_use_kvm_aia(s))".
Is this solely because riscv_kvm_aplic_request() will be an unused function if the crop happens and, as you said above, "the compiler won't eliminate entire unused functions with -O0"?
Thanks,
Daniel
Re: [PATCH 2/2] hw/riscv/virt.c: fix non-KVM --enable-debug build
Quoted 31 lines · Daniel Henrique Barboza
TBH I'm bothered why this doesn't work:
hw/intc/riscv_aplic.c +9
-5@@ -839,12 +839,16 @@ static void riscv_aplic_realize(DeviceState *dev, Error **errp)* Only root APLICs have hardware IRQ lines. All non-root APLICs * have IRQ lines delegated by their parent APLIC. */ - if (!aplic->parent) { - if (is_kvm_aia(aplic->msimode)) { - qdev_init_gpio_in(dev, riscv_kvm_aplic_request, aplic->num_irqs); - } else { - qdev_init_gpio_in(dev, riscv_aplic_request, aplic->num_irqs); + if (kvm_enabled()) { + if (!aplic->parent) { + if (is_kvm_aia(aplic->msimode)) { + qdev_init_gpio_in(dev, riscv_kvm_aplic_request, aplic->num_irqs); + } else { + qdev_init_gpio_in(dev, riscv_aplic_request, aplic->num_irqs); + } } + } else if (!aplic->parent) { + qdev_init_gpio_in(dev, riscv_aplic_request, aplic->num_irqs); }Why is the compiler refusing to crop an "if
kvm_enabled()" block? There's no other conditionals to handle, and it is able to crop "if (kvm_enabled()&&virt_use_kvm_aia(s))".
Quoted 3 lines
Is this solely because
riscv_kvm_aplic_request()will be an unused function if the crop happens and, as you said above, "the compiler won't eliminate entire unused functions with -O0"?
Yes, exactly.
r~