| Message ID | 1627982871-29807-1-git-send-email-vincent.chen@sifive.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] riscv: Ensure the value of FP registers in the core dump file is up to date | expand |
On Tue, 3 Aug 2021 17:27:51 +0800 Vincent Chen <vincent.chen@sifive.com> wrote: > The value of FP registers in the core dump file comes from the > thread.fstate. However, kernel saves the FP registers to the thread.fstate > only before scheduling out the process. If no process switch happens > during the exception handling process, kernel will not have a chance to > save the latest value of FP registers to thread.fstate. It will cause the > value of FP registers in the core dump file may be incorrect. To solve this > problem, this patch force lets kernel save the FP register into the > thread.fstate if the target task_struct equals the current. Nice catch! I think there's another case too: since last kernel save, the FP registers has been modified by the userspace, we also need to force fstate_save() This patch fixes both cases. > > Signed-off-by: Vincent Chen <vincent.chen@sifive.com> Reviewed-by: Jisheng Zhang <jszhang@kernel.org> > > --- > Changes since the v1 patch > 1. Include switch_to.h to avoid compiling error. > > --- > arch/riscv/kernel/ptrace.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/arch/riscv/kernel/ptrace.c b/arch/riscv/kernel/ptrace.c > index 1a85305720e8..9c0511119bad 100644 > --- a/arch/riscv/kernel/ptrace.c > +++ b/arch/riscv/kernel/ptrace.c > @@ -10,6 +10,7 @@ > #include <asm/ptrace.h> > #include <asm/syscall.h> > #include <asm/thread_info.h> > +#include <asm/switch_to.h> > #include <linux/audit.h> > #include <linux/ptrace.h> > #include <linux/elf.h> > @@ -56,6 +57,9 @@ static int riscv_fpr_get(struct task_struct *target, > { > struct __riscv_d_ext_state *fstate = &target->thread.fstate; > > + if (target == current) > + fstate_save(current, task_pt_regs(current)); > + > membuf_write(&to, fstate, offsetof(struct __riscv_d_ext_state, fcsr)); > membuf_store(&to, fstate->fcsr); > return membuf_zero(&to, 4); // explicitly pad
On Tue, 03 Aug 2021 02:27:51 PDT (-0700), vincent.chen@sifive.com wrote: > The value of FP registers in the core dump file comes from the > thread.fstate. However, kernel saves the FP registers to the thread.fstate > only before scheduling out the process. If no process switch happens > during the exception handling process, kernel will not have a chance to > save the latest value of FP registers to thread.fstate. It will cause the > value of FP registers in the core dump file may be incorrect. To solve this > problem, this patch force lets kernel save the FP register into the > thread.fstate if the target task_struct equals the current. > > Signed-off-by: Vincent Chen <vincent.chen@sifive.com> > > --- > Changes since the v1 patch > 1. Include switch_to.h to avoid compiling error. > > --- > arch/riscv/kernel/ptrace.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/arch/riscv/kernel/ptrace.c b/arch/riscv/kernel/ptrace.c > index 1a85305720e8..9c0511119bad 100644 > --- a/arch/riscv/kernel/ptrace.c > +++ b/arch/riscv/kernel/ptrace.c > @@ -10,6 +10,7 @@ > #include <asm/ptrace.h> > #include <asm/syscall.h> > #include <asm/thread_info.h> > +#include <asm/switch_to.h> > #include <linux/audit.h> > #include <linux/ptrace.h> > #include <linux/elf.h> > @@ -56,6 +57,9 @@ static int riscv_fpr_get(struct task_struct *target, > { > struct __riscv_d_ext_state *fstate = &target->thread.fstate; > > + if (target == current) > + fstate_save(current, task_pt_regs(current)); > + > membuf_write(&to, fstate, offsetof(struct __riscv_d_ext_state, fcsr)); > membuf_store(&to, fstate->fcsr); > return membuf_zero(&to, 4); // explicitly pad Thanks, this is on fixes.
diff --git a/arch/riscv/kernel/ptrace.c b/arch/riscv/kernel/ptrace.c index 1a85305720e8..9c0511119bad 100644 --- a/arch/riscv/kernel/ptrace.c +++ b/arch/riscv/kernel/ptrace.c @@ -10,6 +10,7 @@ #include <asm/ptrace.h> #include <asm/syscall.h> #include <asm/thread_info.h> +#include <asm/switch_to.h> #include <linux/audit.h> #include <linux/ptrace.h> #include <linux/elf.h> @@ -56,6 +57,9 @@ static int riscv_fpr_get(struct task_struct *target, { struct __riscv_d_ext_state *fstate = &target->thread.fstate; + if (target == current) + fstate_save(current, task_pt_regs(current)); + membuf_write(&to, fstate, offsetof(struct __riscv_d_ext_state, fcsr)); membuf_store(&to, fstate->fcsr); return membuf_zero(&to, 4); // explicitly pad
The value of FP registers in the core dump file comes from the thread.fstate. However, kernel saves the FP registers to the thread.fstate only before scheduling out the process. If no process switch happens during the exception handling process, kernel will not have a chance to save the latest value of FP registers to thread.fstate. It will cause the value of FP registers in the core dump file may be incorrect. To solve this problem, this patch force lets kernel save the FP register into the thread.fstate if the target task_struct equals the current. Signed-off-by: Vincent Chen <vincent.chen@sifive.com> --- Changes since the v1 patch 1. Include switch_to.h to avoid compiling error. --- arch/riscv/kernel/ptrace.c | 4 ++++ 1 file changed, 4 insertions(+)