| Message ID | 20190802170300.20662-12-cristian.marussi@arm.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Add arm64/signal initial kselftest support | expand |
On 02/08/2019 18:03, Cristian Marussi wrote: > Added a simple fake_sigreturn testcase which places a valid > sigframe on a non-16 bytes aligned SP. > fake_sigretrun() helper function has been patched accordingly > to support placing a sigframe on a non-16 bytes aligned address. > Expects a SIGSEGV on test PASS. > > Adds also a test TODO lists holding some further test ideas. > > Signed-off-by: Cristian Marussi <cristian.marussi@arm.com> > --- > Re-added this text after fixing the forced misaglinment procedure in > fake_sigreturn() itself: require a ZERO alignment and you'll get > your sigframe placed on a misaligned SP (2-bytes off the 16-align) > --- > .../testing/selftests/arm64/signal/signals.S | 21 +++++++++---- > .../arm64/signal/testcases/TODO.readme | 8 +++++ > .../testcases/fake_sigreturn_misaligned_sp.c | 30 +++++++++++++++++++ > 3 files changed, 53 insertions(+), 6 deletions(-) > create mode 100644 tools/testing/selftests/arm64/signal/testcases/TODO.readme > create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > When this test was re-added in V3, the related .gitignore was missed. It will go in V4 Cheers Cristian > diff --git a/tools/testing/selftests/arm64/signal/signals.S b/tools/testing/selftests/arm64/signal/signals.S > index 6262b877400b..2099871176ed 100644 > --- a/tools/testing/selftests/arm64/signal/signals.S > +++ b/tools/testing/selftests/arm64/signal/signals.S > @@ -13,19 +13,28 @@ call_fmt: > > .globl fake_sigreturn > > -/* fake_sigreturn x0:&sigframe, x1:sigframe_size, x2:alignment_SP */ > +/* fake_sigreturn x0:&sigframe, x1:sigframe_sz, x2:align */ > fake_sigreturn: > - mov x20, x0 > - mov x21, x1 > - mov x22, x2 > - mov x23, sp > > - /* create space on the stack for fake sigframe..."x22"-aligned */ > + /* Save args and decide which aligment to enforce */ > + mov x23, sp > + mov x20, x0 > + mov x21, x1 > + /* x22 and x24 used for forcing alignment or misalignment */ > + mov x22, x2 > + mov x24, #0 > + cbnz x22, 1f > + mov x22, #16 > + mov x24, #2 > + > +1: /* create space on the stack for fake sigframe..."x22"-aligned */ > mov x0, #0 > add x0, x21, x22 > sub x22, x22, #1 > bic x0, x0, x22 > sub x23, x23, x0 > + /* force misaligned by x24 bytes if required alignment was zero */ > + add x23, x23, x24 > > ldr x0, =call_fmt > mov x1, x21 > diff --git a/tools/testing/selftests/arm64/signal/testcases/TODO.readme b/tools/testing/selftests/arm64/signal/testcases/TODO.readme > new file mode 100644 > index 000000000000..5c949492e7ab > --- /dev/null > +++ b/tools/testing/selftests/arm64/signal/testcases/TODO.readme > @@ -0,0 +1,8 @@ > +Some more possible ideas for signals tests: > + > +- fake_sigreturn_unmapped_sp > +- fake_sigreturn_kernelspace_sp > +- fake_sigreturn_sve_bad_extra_context > +- mangle_sve_invalid_extra_context > +- mangle_pstate_invalid_el for H modes (+ macroization ?) > +- fake_sigreturn_overflow_reserved > diff --git a/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > new file mode 100644 > index 000000000000..3ee8c500c7d1 > --- /dev/null > +++ b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > @@ -0,0 +1,30 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +/* Copyright (C) 2019 ARM Limited */ > + > +#include <ucontext.h> > + > +#include "test_signals_utils.h" > +#include "testcases.h" > + > +struct fake_sigframe sf; > + > +static int fake_sigreturn_misaligned_run(struct tdescr *td, > + siginfo_t *si, ucontext_t *uc) > +{ > + /* just to fill the ucontext_t with something real */ > + if (!get_current_context(td, &sf.uc)) > + return 1; > + > + /* Forcing sigframe on misaligned (=!16) SP */ > + fake_sigreturn(&sf, sizeof(sf), 0); > + > + return 1; > +} > + > +struct tdescr tde = { > + .name = "FAKE_SIGRETURN_MISALIGNED_SP", > + .descr = "Triggers a fake sigreturn with a misaligned sigframe on SP", > + .sig_ok = SIGSEGV, > + .timeout = 3, > + .run = fake_sigreturn_misaligned_run, > +}; >
On Fri, Aug 02, 2019 at 06:03:00PM +0100, Cristian Marussi wrote: > Added a simple fake_sigreturn testcase which places a valid Add > sigframe on a non-16 bytes aligned SP. > fake_sigretrun() helper function has been patched accordingly > to support placing a sigframe on a non-16 bytes aligned address. > Expects a SIGSEGV on test PASS. > > Adds also a test TODO lists holding some further test ideas. Adds -> Also add lists -> list > Signed-off-by: Cristian Marussi <cristian.marussi@arm.com> > --- > Re-added this text after fixing the forced misaglinment procedure in > fake_sigreturn() itself: require a ZERO alignment and you'll get > your sigframe placed on a misaligned SP (2-bytes off the 16-align) > --- > .../testing/selftests/arm64/signal/signals.S | 21 +++++++++---- > .../arm64/signal/testcases/TODO.readme | 8 +++++ > .../testcases/fake_sigreturn_misaligned_sp.c | 30 +++++++++++++++++++ > 3 files changed, 53 insertions(+), 6 deletions(-) > create mode 100644 tools/testing/selftests/arm64/signal/testcases/TODO.readme > create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > > diff --git a/tools/testing/selftests/arm64/signal/signals.S b/tools/testing/selftests/arm64/signal/signals.S > index 6262b877400b..2099871176ed 100644 > --- a/tools/testing/selftests/arm64/signal/signals.S > +++ b/tools/testing/selftests/arm64/signal/signals.S > @@ -13,19 +13,28 @@ call_fmt: > > .globl fake_sigreturn > > -/* fake_sigreturn x0:&sigframe, x1:sigframe_size, x2:alignment_SP */ > +/* fake_sigreturn x0:&sigframe, x1:sigframe_sz, x2:align */ > fake_sigreturn: > - mov x20, x0 > - mov x21, x1 > - mov x22, x2 > - mov x23, sp > > - /* create space on the stack for fake sigframe..."x22"-aligned */ > + /* Save args and decide which aligment to enforce */ > + mov x23, sp > + mov x20, x0 > + mov x21, x1 > + /* x22 and x24 used for forcing alignment or misalignment */ > + mov x22, x2 > + mov x24, #0 > + cbnz x22, 1f > + mov x22, #16 > + mov x24, #2 > + > +1: /* create space on the stack for fake sigframe..."x22"-aligned */ > mov x0, #0 > add x0, x21, x22 > sub x22, x22, #1 > bic x0, x0, x22 > sub x23, x23, x0 > + /* force misaligned by x24 bytes if required alignment was zero */ > + add x23, x23, x24 > > ldr x0, =call_fmt > mov x1, x21 Would it be simpler for the third argument to specify a number of bytes to subtract from SP after allocating 16-byte aligned storage to accommodate sigframe_sz? Then 0 gives an aligned frame, 1 gives a frame misaligned by 1 byte, etc. Also if all this is a fix to the original fake_sigreturn, can we merge it into the original patch instead? > diff --git a/tools/testing/selftests/arm64/signal/testcases/TODO.readme b/tools/testing/selftests/arm64/signal/testcases/TODO.readme > new file mode 100644 > index 000000000000..5c949492e7ab > --- /dev/null > +++ b/tools/testing/selftests/arm64/signal/testcases/TODO.readme > @@ -0,0 +1,8 @@ > +Some more possible ideas for signals tests: > + > +- fake_sigreturn_unmapped_sp > +- fake_sigreturn_kernelspace_sp > +- fake_sigreturn_sve_bad_extra_context > +- mangle_sve_invalid_extra_context > +- mangle_pstate_invalid_el for H modes (+ macroization ?) > +- fake_sigreturn_overflow_reserved This seems a reasonable list, but it occurs to me that it will tend to go out of sync as tests get added. So maybe just put this list in the cover letter instead of including it in the patch. We should probably have a one-line description of each proposed test, since the names are a bit cryptic. > diff --git a/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > new file mode 100644 > index 000000000000..3ee8c500c7d1 > --- /dev/null > +++ b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > @@ -0,0 +1,30 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +/* Copyright (C) 2019 ARM Limited */ > + signal.h? > +#include <ucontext.h> > + > +#include "test_signals_utils.h" > +#include "testcases.h" > + > +struct fake_sigframe sf; > + > +static int fake_sigreturn_misaligned_run(struct tdescr *td, > + siginfo_t *si, ucontext_t *uc) > +{ > + /* just to fill the ucontext_t with something real */ > + if (!get_current_context(td, &sf.uc)) > + return 1; > + > + /* Forcing sigframe on misaligned (=!16) SP */ > + fake_sigreturn(&sf, sizeof(sf), 0); > + > + return 1; > +} [...] Cheers ---Dave
On Wed, Aug 07, 2019 at 05:04:13PM +0100, Cristian Marussi wrote: > On 02/08/2019 18:03, Cristian Marussi wrote: > > Added a simple fake_sigreturn testcase which places a valid > > sigframe on a non-16 bytes aligned SP. > > fake_sigretrun() helper function has been patched accordingly > > to support placing a sigframe on a non-16 bytes aligned address. > > Expects a SIGSEGV on test PASS. > > > > Adds also a test TODO lists holding some further test ideas. > > > > Signed-off-by: Cristian Marussi <cristian.marussi@arm.com> > > --- > > Re-added this text after fixing the forced misaglinment procedure in > > fake_sigreturn() itself: require a ZERO alignment and you'll get > > your sigframe placed on a misaligned SP (2-bytes off the 16-align) > > --- > > .../testing/selftests/arm64/signal/signals.S | 21 +++++++++---- > > .../arm64/signal/testcases/TODO.readme | 8 +++++ > > .../testcases/fake_sigreturn_misaligned_sp.c | 30 +++++++++++++++++++ > > 3 files changed, 53 insertions(+), 6 deletions(-) > > create mode 100644 tools/testing/selftests/arm64/signal/testcases/TODO.readme > > create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c > > > > When this test was re-added in V3, the related .gitignore was missed. > It will go in V4 Ack, or otherwise try switching to using wildcards in .gitignore as suggested in my reply to patch 4. [...] Cheers ---Dave
On 13/08/2019 17:28, Dave Martin wrote: > On Wed, Aug 07, 2019 at 05:04:13PM +0100, Cristian Marussi wrote: >> On 02/08/2019 18:03, Cristian Marussi wrote: >>> Added a simple fake_sigreturn testcase which places a valid >>> sigframe on a non-16 bytes aligned SP. >>> fake_sigretrun() helper function has been patched accordingly >>> to support placing a sigframe on a non-16 bytes aligned address. >>> Expects a SIGSEGV on test PASS. >>> >>> Adds also a test TODO lists holding some further test ideas. >>> >>> Signed-off-by: Cristian Marussi <cristian.marussi@arm.com> >>> --- >>> Re-added this text after fixing the forced misaglinment procedure in >>> fake_sigreturn() itself: require a ZERO alignment and you'll get >>> your sigframe placed on a misaligned SP (2-bytes off the 16-align) >>> --- >>> .../testing/selftests/arm64/signal/signals.S | 21 +++++++++---- >>> .../arm64/signal/testcases/TODO.readme | 8 +++++ >>> .../testcases/fake_sigreturn_misaligned_sp.c | 30 +++++++++++++++++++ >>> 3 files changed, 53 insertions(+), 6 deletions(-) >>> create mode 100644 tools/testing/selftests/arm64/signal/testcases/TODO.readme >>> create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c >>> >> >> When this test was re-added in V3, the related .gitignore was missed. >> It will go in V4 > > Ack, or otherwise try switching to using wildcards in .gitignore as > suggested in my reply to patch 4. Used wildcard in .gitignore as advised. > > [...] > > Cheers > ---Dave > Cheers Cristian
On 13/08/2019 17:27, Dave Martin wrote: > On Fri, Aug 02, 2019 at 06:03:00PM +0100, Cristian Marussi wrote: >> Added a simple fake_sigreturn testcase which places a valid > > Add > Ok >> sigframe on a non-16 bytes aligned SP. >> fake_sigretrun() helper function has been patched accordingly >> to support placing a sigframe on a non-16 bytes aligned address. >> Expects a SIGSEGV on test PASS. >> >> Adds also a test TODO lists holding some further test ideas. > > Adds -> Also add > lists -> list > Removed together with the TODO. >> Signed-off-by: Cristian Marussi <cristian.marussi@arm.com> >> --- >> Re-added this text after fixing the forced misaglinment procedure in >> fake_sigreturn() itself: require a ZERO alignment and you'll get >> your sigframe placed on a misaligned SP (2-bytes off the 16-align) >> --- >> .../testing/selftests/arm64/signal/signals.S | 21 +++++++++---- >> .../arm64/signal/testcases/TODO.readme | 8 +++++ >> .../testcases/fake_sigreturn_misaligned_sp.c | 30 +++++++++++++++++++ >> 3 files changed, 53 insertions(+), 6 deletions(-) >> create mode 100644 tools/testing/selftests/arm64/signal/testcases/TODO.readme >> create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c >> >> diff --git a/tools/testing/selftests/arm64/signal/signals.S b/tools/testing/selftests/arm64/signal/signals.S >> index 6262b877400b..2099871176ed 100644 >> --- a/tools/testing/selftests/arm64/signal/signals.S >> +++ b/tools/testing/selftests/arm64/signal/signals.S >> @@ -13,19 +13,28 @@ call_fmt: >> >> .globl fake_sigreturn >> >> -/* fake_sigreturn x0:&sigframe, x1:sigframe_size, x2:alignment_SP */ >> +/* fake_sigreturn x0:&sigframe, x1:sigframe_sz, x2:align */ >> fake_sigreturn: >> - mov x20, x0 >> - mov x21, x1 >> - mov x22, x2 >> - mov x23, sp >> >> - /* create space on the stack for fake sigframe..."x22"-aligned */ >> + /* Save args and decide which aligment to enforce */ >> + mov x23, sp >> + mov x20, x0 >> + mov x21, x1 >> + /* x22 and x24 used for forcing alignment or misalignment */ >> + mov x22, x2 >> + mov x24, #0 >> + cbnz x22, 1f >> + mov x22, #16 >> + mov x24, #2 >> + >> +1: /* create space on the stack for fake sigframe..."x22"-aligned */ >> mov x0, #0 >> add x0, x21, x22 >> sub x22, x22, #1 >> bic x0, x0, x22 >> sub x23, x23, x0 >> + /* force misaligned by x24 bytes if required alignment was zero */ >> + add x23, x23, x24 >> >> ldr x0, =call_fmt >> mov x1, x21 > > Would it be simpler for the third argument to specify a number of bytes > to subtract from SP after allocating 16-byte aligned storage to > accommodate sigframe_sz? > > Then 0 gives an aligned frame, 1 gives a frame misaligned by 1 byte, > etc. > > > Also if all this is a fix to the original fake_sigreturn, can we merge > it into the original patch instead? > Yes simplified and merged into 6/11 >> diff --git a/tools/testing/selftests/arm64/signal/testcases/TODO.readme b/tools/testing/selftests/arm64/signal/testcases/TODO.readme >> new file mode 100644 >> index 000000000000..5c949492e7ab >> --- /dev/null >> +++ b/tools/testing/selftests/arm64/signal/testcases/TODO.readme >> @@ -0,0 +1,8 @@ >> +Some more possible ideas for signals tests: >> + >> +- fake_sigreturn_unmapped_sp >> +- fake_sigreturn_kernelspace_sp >> +- fake_sigreturn_sve_bad_extra_context >> +- mangle_sve_invalid_extra_context >> +- mangle_pstate_invalid_el for H modes (+ macroization ?) >> +- fake_sigreturn_overflow_reserved > > This seems a reasonable list, but it occurs to me that it will tend to > go out of sync as tests get added. So maybe just put this list in the > cover letter instead of including it in the patch. > > We should probably have a one-line description of each proposed test, > since the names are a bit cryptic. Moving to cover letter with description. > >> diff --git a/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c >> new file mode 100644 >> index 000000000000..3ee8c500c7d1 >> --- /dev/null >> +++ b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c >> @@ -0,0 +1,30 @@ >> +/* SPDX-License-Identifier: GPL-2.0 */ >> +/* Copyright (C) 2019 ARM Limited */ >> + > > signal.h? ok > >> +#include <ucontext.h> >> + >> +#include "test_signals_utils.h" >> +#include "testcases.h" >> + >> +struct fake_sigframe sf; >> + >> +static int fake_sigreturn_misaligned_run(struct tdescr *td, >> + siginfo_t *si, ucontext_t *uc) >> +{ >> + /* just to fill the ucontext_t with something real */ >> + if (!get_current_context(td, &sf.uc)) >> + return 1; >> + >> + /* Forcing sigframe on misaligned (=!16) SP */ >> + fake_sigreturn(&sf, sizeof(sf), 0); >> + >> + return 1; >> +} > > [...] > > Cheers > ---Dave > Cheers Cristian
diff --git a/tools/testing/selftests/arm64/signal/signals.S b/tools/testing/selftests/arm64/signal/signals.S index 6262b877400b..2099871176ed 100644 --- a/tools/testing/selftests/arm64/signal/signals.S +++ b/tools/testing/selftests/arm64/signal/signals.S @@ -13,19 +13,28 @@ call_fmt: .globl fake_sigreturn -/* fake_sigreturn x0:&sigframe, x1:sigframe_size, x2:alignment_SP */ +/* fake_sigreturn x0:&sigframe, x1:sigframe_sz, x2:align */ fake_sigreturn: - mov x20, x0 - mov x21, x1 - mov x22, x2 - mov x23, sp - /* create space on the stack for fake sigframe..."x22"-aligned */ + /* Save args and decide which aligment to enforce */ + mov x23, sp + mov x20, x0 + mov x21, x1 + /* x22 and x24 used for forcing alignment or misalignment */ + mov x22, x2 + mov x24, #0 + cbnz x22, 1f + mov x22, #16 + mov x24, #2 + +1: /* create space on the stack for fake sigframe..."x22"-aligned */ mov x0, #0 add x0, x21, x22 sub x22, x22, #1 bic x0, x0, x22 sub x23, x23, x0 + /* force misaligned by x24 bytes if required alignment was zero */ + add x23, x23, x24 ldr x0, =call_fmt mov x1, x21 diff --git a/tools/testing/selftests/arm64/signal/testcases/TODO.readme b/tools/testing/selftests/arm64/signal/testcases/TODO.readme new file mode 100644 index 000000000000..5c949492e7ab --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/TODO.readme @@ -0,0 +1,8 @@ +Some more possible ideas for signals tests: + +- fake_sigreturn_unmapped_sp +- fake_sigreturn_kernelspace_sp +- fake_sigreturn_sve_bad_extra_context +- mangle_sve_invalid_extra_context +- mangle_pstate_invalid_el for H modes (+ macroization ?) +- fake_sigreturn_overflow_reserved diff --git a/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c new file mode 100644 index 000000000000..3ee8c500c7d1 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c @@ -0,0 +1,30 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright (C) 2019 ARM Limited */ + +#include <ucontext.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; + +static int fake_sigreturn_misaligned_run(struct tdescr *td, + siginfo_t *si, ucontext_t *uc) +{ + /* just to fill the ucontext_t with something real */ + if (!get_current_context(td, &sf.uc)) + return 1; + + /* Forcing sigframe on misaligned (=!16) SP */ + fake_sigreturn(&sf, sizeof(sf), 0); + + return 1; +} + +struct tdescr tde = { + .name = "FAKE_SIGRETURN_MISALIGNED_SP", + .descr = "Triggers a fake sigreturn with a misaligned sigframe on SP", + .sig_ok = SIGSEGV, + .timeout = 3, + .run = fake_sigreturn_misaligned_run, +};
Added a simple fake_sigreturn testcase which places a valid sigframe on a non-16 bytes aligned SP. fake_sigretrun() helper function has been patched accordingly to support placing a sigframe on a non-16 bytes aligned address. Expects a SIGSEGV on test PASS. Adds also a test TODO lists holding some further test ideas. Signed-off-by: Cristian Marussi <cristian.marussi@arm.com> --- Re-added this text after fixing the forced misaglinment procedure in fake_sigreturn() itself: require a ZERO alignment and you'll get your sigframe placed on a misaligned SP (2-bytes off the 16-align) --- .../testing/selftests/arm64/signal/signals.S | 21 +++++++++---- .../arm64/signal/testcases/TODO.readme | 8 +++++ .../testcases/fake_sigreturn_misaligned_sp.c | 30 +++++++++++++++++++ 3 files changed, 53 insertions(+), 6 deletions(-) create mode 100644 tools/testing/selftests/arm64/signal/testcases/TODO.readme create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_misaligned_sp.c