Message ID | 20180413142336.32163-1-peter.maydell@linaro.org (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
On Fri, Apr 13, 2018 at 03:23:36PM +0100, Peter Maydell wrote: > The MIPS TCG target makes the assumption that the offset from the > target env pointer to the tlb_table is less than about 64K. This > used to be true, but gradual addition of features to the Arm > target means that it's no longer true there. This results in > the build-time assertion failing: > > In file included from /home/pm215/qemu/include/qemu/osdep.h:36:0, > from /home/pm215/qemu/tcg/tcg.c:28: > /home/pm215/qemu/tcg/mips/tcg-target.inc.c: In function ‘tcg_out_tlb_load’: > /home/pm215/qemu/include/qemu/compiler.h:90:36: error: static assertion failed: "not expecting: offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) > 0x7ff0 + 0x7fff" > #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg) > ^ > /home/pm215/qemu/include/qemu/compiler.h:98:30: note: in expansion of macro ‘QEMU_BUILD_BUG_MSG’ > #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x) > ^ > /home/pm215/qemu/tcg/mips/tcg-target.inc.c:1236:9: note: in expansion of macro ‘QEMU_BUILD_BUG_ON’ > QEMU_BUILD_BUG_ON(offsetof(CPUArchState, > ^ > /home/pm215/qemu/rules.mak:66: recipe for target 'tcg/tcg.o' failed > > An ideal long term approach would be to rearrange the CPU state > so that the tlb_table was not so far along it, but this is tricky > because it would move it from the "not cleared on CPU reset" part > of the struct to the "cleared on CPU reset" part. As a simple fix > for the 2.12 release, make the MIPS TCG target handle an arbitrary > offset by emitting more add instructions. This will mean an extra > instruction in the fastpath for TCG loads and stores for the > affected guests (currently just aarch64-softmmu). > > Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Acked-by: Michael S. Tsirkin <mst@redhat.com> > --- > This is sufficient that on MIPS host we can now at least build > and run an aarch64 guest kernel. I haven't tried 'make check' > because the only MIPS system I have access to is way too slow... > > tcg/mips/tcg-target.inc.c | 11 ++++------- > 1 file changed, 4 insertions(+), 7 deletions(-) > > diff --git a/tcg/mips/tcg-target.inc.c b/tcg/mips/tcg-target.inc.c > index 4b55ab8856..ca5f1d4894 100644 > --- a/tcg/mips/tcg-target.inc.c > +++ b/tcg/mips/tcg-target.inc.c > @@ -1229,13 +1229,10 @@ static void tcg_out_tlb_load(TCGContext *s, TCGReg base, TCGReg addrl, > tcg_out_opc_reg(s, ALIAS_PADD, TCG_REG_A0, TCG_REG_A0, TCG_AREG0); > > /* Compensate for very large offsets. */ > - if (add_off >= 0x8000) { > - /* Most target env are smaller than 32k; none are larger than 64k. > - Simplify the logic here merely to offset by 0x7ff0, giving us a > - range just shy of 64k. Check this assumption. */ > - QEMU_BUILD_BUG_ON(offsetof(CPUArchState, > - tlb_table[NB_MMU_MODES - 1][1]) > - > 0x7ff0 + 0x7fff); > + while (add_off >= 0x8000) { > + /* Most target env are smaller than 32k, but a few are larger than 64k, > + * so handle an arbitrarily large offset. > + */ > tcg_out_opc_imm(s, ALIAS_PADDI, TCG_REG_A0, TCG_REG_A0, 0x7ff0); > cmp_off -= 0x7ff0; > add_off -= 0x7ff0; > -- > 2.16.2
On 04/13/2018 04:23 AM, Peter Maydell wrote: > The MIPS TCG target makes the assumption that the offset from the > target env pointer to the tlb_table is less than about 64K. This > used to be true, but gradual addition of features to the Arm > target means that it's no longer true there. This results in > the build-time assertion failing: > > In file included from /home/pm215/qemu/include/qemu/osdep.h:36:0, > from /home/pm215/qemu/tcg/tcg.c:28: > /home/pm215/qemu/tcg/mips/tcg-target.inc.c: In function ‘tcg_out_tlb_load’: > /home/pm215/qemu/include/qemu/compiler.h:90:36: error: static assertion failed: "not expecting: offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) > 0x7ff0 + 0x7fff" > #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg) > ^ > /home/pm215/qemu/include/qemu/compiler.h:98:30: note: in expansion of macro ‘QEMU_BUILD_BUG_MSG’ > #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x) > ^ > /home/pm215/qemu/tcg/mips/tcg-target.inc.c:1236:9: note: in expansion of macro ‘QEMU_BUILD_BUG_ON’ > QEMU_BUILD_BUG_ON(offsetof(CPUArchState, > ^ > /home/pm215/qemu/rules.mak:66: recipe for target 'tcg/tcg.o' failed Yes, I asked for help on this from the MIPS folks back in January when I posted patches for ppc and arm(32) hosts. And then of course forgot about it in the interim. > + while (add_off >= 0x8000) { > + /* Most target env are smaller than 32k, but a few are larger than 64k, > + * so handle an arbitrarily large offset. > + */ > tcg_out_opc_imm(s, ALIAS_PADDI, TCG_REG_A0, TCG_REG_A0, 0x7ff0); > cmp_off -= 0x7ff0; > add_off -= 0x7ff0; This is a pretty darned good solution, really. I should have thought of it myself at the time. The new AArch64 offset is about 80k, so there will be two adds emitted. Loading C<<16 into a temp register and then adding would be no better in the end (one can only add C<<16 directly with mipsr6 extensions). Reviewed-by: Richard Henderson <richard.henderson@linaro.org> r~
On 13 April 2018 at 20:09, Richard Henderson <rth@twiddle.net> wrote: > On 04/13/2018 04:23 AM, Peter Maydell wrote: >> The MIPS TCG target makes the assumption that the offset from the >> target env pointer to the tlb_table is less than about 64K. This >> used to be true, but gradual addition of features to the Arm >> target means that it's no longer true there. This results in >> the build-time assertion failing: >> >> In file included from /home/pm215/qemu/include/qemu/osdep.h:36:0, >> from /home/pm215/qemu/tcg/tcg.c:28: >> /home/pm215/qemu/tcg/mips/tcg-target.inc.c: In function ‘tcg_out_tlb_load’: >> /home/pm215/qemu/include/qemu/compiler.h:90:36: error: static assertion failed: "not expecting: offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) > 0x7ff0 + 0x7fff" >> #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg) >> ^ >> /home/pm215/qemu/include/qemu/compiler.h:98:30: note: in expansion of macro ‘QEMU_BUILD_BUG_MSG’ >> #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x) >> ^ >> /home/pm215/qemu/tcg/mips/tcg-target.inc.c:1236:9: note: in expansion of macro ‘QEMU_BUILD_BUG_ON’ >> QEMU_BUILD_BUG_ON(offsetof(CPUArchState, >> ^ >> /home/pm215/qemu/rules.mak:66: recipe for target 'tcg/tcg.o' failed > > Yes, I asked for help on this from the MIPS folks back in January when I posted > patches for ppc and arm(32) hosts. And then of course forgot about it in the > interim. > >> + while (add_off >= 0x8000) { >> + /* Most target env are smaller than 32k, but a few are larger than 64k, >> + * so handle an arbitrarily large offset. >> + */ >> tcg_out_opc_imm(s, ALIAS_PADDI, TCG_REG_A0, TCG_REG_A0, 0x7ff0); >> cmp_off -= 0x7ff0; >> add_off -= 0x7ff0; > > This is a pretty darned good solution, really. I should have thought of it > myself at the time. The new AArch64 offset is about 80k, so there will be two > adds emitted. Loading C<<16 into a temp register and then adding would be no > better in the end (one can only add C<<16 directly with mipsr6 extensions). > > Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Thanks; applied to master. -- PMM
Peter Maydell <peter.maydell@linaro.org> writes: > The MIPS TCG target makes the assumption that the offset from the > target env pointer to the tlb_table is less than about 64K. This > used to be true, but gradual addition of features to the Arm > target means that it's no longer true there. This results in > the build-time assertion failing: > > In file included from /home/pm215/qemu/include/qemu/osdep.h:36:0, > from /home/pm215/qemu/tcg/tcg.c:28: > /home/pm215/qemu/tcg/mips/tcg-target.inc.c: In function ‘tcg_out_tlb_load’: > /home/pm215/qemu/include/qemu/compiler.h:90:36: error: static assertion failed: "not expecting: offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) > 0x7ff0 + 0x7fff" > #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg) > ^ > /home/pm215/qemu/include/qemu/compiler.h:98:30: note: in expansion of macro ‘QEMU_BUILD_BUG_MSG’ > #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x) > ^ > /home/pm215/qemu/tcg/mips/tcg-target.inc.c:1236:9: note: in expansion of macro ‘QEMU_BUILD_BUG_ON’ > QEMU_BUILD_BUG_ON(offsetof(CPUArchState, > ^ > /home/pm215/qemu/rules.mak:66: recipe for target 'tcg/tcg.o' failed > > An ideal long term approach would be to rearrange the CPU state > so that the tlb_table was not so far along it, but this is tricky > because it would move it from the "not cleared on CPU reset" part > of the struct to the "cleared on CPU reset" part. Is that really a problem? Doesn't it mean we'll just reload the TLB after a reset? -- Alex Bennée
On 30 April 2018 at 18:44, Alex Bennée <alex.bennee@linaro.org> wrote: > Peter Maydell <peter.maydell@linaro.org> writes: >> An ideal long term approach would be to rearrange the CPU state >> so that the tlb_table was not so far along it, but this is tricky >> because it would move it from the "not cleared on CPU reset" part >> of the struct to the "cleared on CPU reset" part. > > Is that really a problem? Doesn't it mean we'll just reload the TLB > after a reset? It was the sort of "this subtly changes the semantics on all hosts for the sake of an issue that's mips-hosts-only" change that I didn't want to try to reason through and make very close to release when we didn't have a lot of time to test the results... thanks -- PMM
diff --git a/tcg/mips/tcg-target.inc.c b/tcg/mips/tcg-target.inc.c index 4b55ab8856..ca5f1d4894 100644 --- a/tcg/mips/tcg-target.inc.c +++ b/tcg/mips/tcg-target.inc.c @@ -1229,13 +1229,10 @@ static void tcg_out_tlb_load(TCGContext *s, TCGReg base, TCGReg addrl, tcg_out_opc_reg(s, ALIAS_PADD, TCG_REG_A0, TCG_REG_A0, TCG_AREG0); /* Compensate for very large offsets. */ - if (add_off >= 0x8000) { - /* Most target env are smaller than 32k; none are larger than 64k. - Simplify the logic here merely to offset by 0x7ff0, giving us a - range just shy of 64k. Check this assumption. */ - QEMU_BUILD_BUG_ON(offsetof(CPUArchState, - tlb_table[NB_MMU_MODES - 1][1]) - > 0x7ff0 + 0x7fff); + while (add_off >= 0x8000) { + /* Most target env are smaller than 32k, but a few are larger than 64k, + * so handle an arbitrarily large offset. + */ tcg_out_opc_imm(s, ALIAS_PADDI, TCG_REG_A0, TCG_REG_A0, 0x7ff0); cmp_off -= 0x7ff0; add_off -= 0x7ff0;
The MIPS TCG target makes the assumption that the offset from the target env pointer to the tlb_table is less than about 64K. This used to be true, but gradual addition of features to the Arm target means that it's no longer true there. This results in the build-time assertion failing: In file included from /home/pm215/qemu/include/qemu/osdep.h:36:0, from /home/pm215/qemu/tcg/tcg.c:28: /home/pm215/qemu/tcg/mips/tcg-target.inc.c: In function ‘tcg_out_tlb_load’: /home/pm215/qemu/include/qemu/compiler.h:90:36: error: static assertion failed: "not expecting: offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) > 0x7ff0 + 0x7fff" #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg) ^ /home/pm215/qemu/include/qemu/compiler.h:98:30: note: in expansion of macro ‘QEMU_BUILD_BUG_MSG’ #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x) ^ /home/pm215/qemu/tcg/mips/tcg-target.inc.c:1236:9: note: in expansion of macro ‘QEMU_BUILD_BUG_ON’ QEMU_BUILD_BUG_ON(offsetof(CPUArchState, ^ /home/pm215/qemu/rules.mak:66: recipe for target 'tcg/tcg.o' failed An ideal long term approach would be to rearrange the CPU state so that the tlb_table was not so far along it, but this is tricky because it would move it from the "not cleared on CPU reset" part of the struct to the "cleared on CPU reset" part. As a simple fix for the 2.12 release, make the MIPS TCG target handle an arbitrary offset by emitting more add instructions. This will mean an extra instruction in the fastpath for TCG loads and stores for the affected guests (currently just aarch64-softmmu). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- This is sufficient that on MIPS host we can now at least build and run an aarch64 guest kernel. I haven't tried 'make check' because the only MIPS system I have access to is way too slow... tcg/mips/tcg-target.inc.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-)