| Message ID | 20220624134415.343417-1-broonie@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] arm64/signal: Clean up SVE/SME feature checking inconsistency | expand |
On Fri, Jun 24, 2022 at 02:44:14PM +0100, Mark Brown wrote: > Currently when restoring signal state we check to see if SVE is supported > in restore_sigframe() but check to see if SVE is supported inside > restore_sve_fpsimd_context(). This makes no real difference since SVE is > always supported in systems with SME but looks a bit untidy and makes > things slightly harder to follow, move the SVE check next to the SME one > in restore_sve_fpsimd_context(). > > Signed-off-by: Mark Brown <broonie@kernel.org> > --- > > v2: > - Add a stub restore_sve_fpsimd_context() to hopefully fix an > allnoconfig issue, I can't reproduce locally. > > arch/arm64/kernel/signal.c | 19 +++++++++++-------- > 1 file changed, 11 insertions(+), 8 deletions(-) > > diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c > index b0980fbb6bc7..6b6a79806e82 100644 > --- a/arch/arm64/kernel/signal.c > +++ b/arch/arm64/kernel/signal.c > @@ -280,6 +280,9 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user) > > vl = task_get_sme_vl(current); > } else { > + if (!system_supports_sve()) > + return -EINVAL; > + > vl = task_get_sve_vl(current); > } > > @@ -342,9 +345,13 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user) > > #else /* ! CONFIG_ARM64_SVE */ > > -/* Turn any non-optimised out attempts to use these into a link error: */ > +static int restore_sve_fpsimd_context(struct user_ctxs *user) > +{ > + return 0; > +} Given that this should never be called, should we return an error instead of 0 (or possibly even WARN/BUG)? Will
diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c index b0980fbb6bc7..6b6a79806e82 100644 --- a/arch/arm64/kernel/signal.c +++ b/arch/arm64/kernel/signal.c @@ -280,6 +280,9 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user) vl = task_get_sme_vl(current); } else { + if (!system_supports_sve()) + return -EINVAL; + vl = task_get_sve_vl(current); } @@ -342,9 +345,13 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user) #else /* ! CONFIG_ARM64_SVE */ -/* Turn any non-optimised out attempts to use these into a link error: */ +static int restore_sve_fpsimd_context(struct user_ctxs *user) +{ + return 0; +} + +/* Turn any non-optimised out attempts to use this into a link error: */ extern int preserve_sve_context(void __user *ctx); -extern int restore_sve_fpsimd_context(struct user_ctxs *user); #endif /* ! CONFIG_ARM64_SVE */ @@ -649,14 +656,10 @@ static int restore_sigframe(struct pt_regs *regs, if (!user.fpsimd) return -EINVAL; - if (user.sve) { - if (!system_supports_sve()) - return -EINVAL; - + if (user.sve) err = restore_sve_fpsimd_context(&user); - } else { + else err = restore_fpsimd_context(user.fpsimd); - } } if (err == 0 && system_supports_sme() && user.za)
Currently when restoring signal state we check to see if SVE is supported in restore_sigframe() but check to see if SVE is supported inside restore_sve_fpsimd_context(). This makes no real difference since SVE is always supported in systems with SME but looks a bit untidy and makes things slightly harder to follow, move the SVE check next to the SME one in restore_sve_fpsimd_context(). Signed-off-by: Mark Brown <broonie@kernel.org> --- v2: - Add a stub restore_sve_fpsimd_context() to hopefully fix an allnoconfig issue, I can't reproduce locally. arch/arm64/kernel/signal.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-)