[PATCH 17/27] um: Store full CSGSFS and SS register from mcontext

Benjamin Berg benjamin at sipsolutions.net
Wed Mar 3 15:55:13 GMT 2021


Doing this allows using registers as retrieved from an mcontext to be
pushed to a process using PTRACE_SETREGS.

It is not entirely clear to me why CSGSFS was masked. Doing so creates
issues when using the mcontext as process state in seccomp and simply
copying the register appears to work perfectly fine for ptrace.

Signed-off-by: Benjamin Berg <benjamin at sipsolutions.net>

---

NOTE: I thought that a good additional test for the register store/restore
functions would be to use the registers as extraced using seccomp during
startup to run using ptrace (i.e. force using_seccomp to 0). However,
while this works great for simple tests cases, I have a scenario where
python suddenly fails to parse bytecode.

This is not an issue in general. But I don't understand why it is
happening and it might mean that there is an unexpected mismatch between
the ptrace and seccomp code paths.
---
 arch/x86/um/os-Linux/mcontext.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/arch/x86/um/os-Linux/mcontext.c b/arch/x86/um/os-Linux/mcontext.c
index 49c3744cac37..81b9d1f9f4e6 100644
--- a/arch/x86/um/os-Linux/mcontext.c
+++ b/arch/x86/um/os-Linux/mcontext.c
@@ -26,7 +26,6 @@ void get_regs_from_mc(struct uml_pt_regs *regs, mcontext_t *mc)
 	COPY(RIP);
 	COPY2(EFLAGS, EFL);
 	COPY2(CS, CSGSFS);
-	regs->gp[CS / sizeof(unsigned long)] &= 0xffff;
-	regs->gp[CS / sizeof(unsigned long)] |= 3;
+	regs->gp[SS / sizeof(unsigned long)] = mc->gregs[REG_CSGSFS] >> 48;
 #endif
 }
-- 
2.29.2




More information about the linux-um mailing list