[PATCH v3 00/11] General cleanups and fixes from SECCOMP patchset

[PATCH v3 00/11] General cleanups and fixes from SECCOMP patchset

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

Hi,

the SECCOMP patchset is not ready yet as I need to change it quite a bit
for the FD passing based security model that was proposed.

Until then, the cleanups and small fixes from the start of the patchset
are still relevant and it could be nice to pull them in anyway. So I am
resending just these.

I purely rebased the patchset. There should be no other differences
between v2 and v3 here.

Benjamin

Benjamin Berg (11):
  um: Drop support for hosts without SYSEMU_SINGLESTEP support
  um: Drop NULL check from start_userspace
  um: Make errors to stop ptraced child fatal during startup
  um: Don't use vfprintf() for os_info()
  um: Do not use printk in SIGWINCH helper thread
  um: Reap winch thread if it fails
  um: Do not use printk in userspace trampoline
  um: Always inline stub functions
  um: Rely on PTRACE_SETREGSET to set FS/GS base registers
  um: Remove unused register save/restore functions
  um: Mark 32bit syscall helpers as clobbering memory

 arch/um/drivers/chan_user.c             |  42 +++++----
 arch/um/drivers/line.c                  |  13 ++-
 arch/um/include/asm/processor-generic.h |   1 -
 arch/um/include/shared/kern_util.h      |   3 +-
 arch/um/include/shared/os.h             |   3 -
 arch/um/include/shared/ptrace_user.h    |  41 ---------
 arch/um/include/shared/registers.h      |   2 -
 arch/um/kernel/process.c                |  12 +--
 arch/um/kernel/ptrace.c                 |   2 -
 arch/um/kernel/signal.c                 |  12 ---
 arch/um/os-Linux/registers.c            |  20 ----
 arch/um/os-Linux/skas/process.c         | 117 ++++++++----------------
 arch/um/os-Linux/start_up.c             | 111 ++++------------------
 arch/um/os-Linux/util.c                 |  19 +++-
 arch/x86/um/asm/elf.h                   |   4 +-
 arch/x86/um/asm/processor_64.h          |   3 -
 arch/x86/um/os-Linux/Makefile           |   1 -
 arch/x86/um/os-Linux/prctl.c            |  12 ---
 arch/x86/um/ptrace_32.c                 |  24 -----
 arch/x86/um/ptrace_64.c                 |  26 ------
 arch/x86/um/shared/sysdep/ptrace_32.h   |   4 -
 arch/x86/um/shared/sysdep/ptrace_user.h |  12 +--
 arch/x86/um/shared/sysdep/stub_32.h     |  39 ++++----
 arch/x86/um/shared/sysdep/stub_64.h     |  17 ++--
 arch/x86/um/syscalls_64.c               |  62 +++----------
 arch/x86/um/tls_64.c                    |   2 +-
 26 files changed, 159 insertions(+), 445 deletions(-)
 delete mode 100644 arch/x86/um/os-Linux/prctl.c

-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 01/11] um: Drop support for hosts without SYSEMU_SINGLESTEP support

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

These features have existed since Linux 2.6.14 and can be considered
widely available at this point. Also drop the backward compatibility
code for PTRACE_SETOPTIONS.

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

----

v2:
 * Continue to define PTRACE_SYSEMU_SINGLESTEP as glibc only added it in
   version 2.27.
---
 arch/um/include/asm/processor-generic.h |  1 -
 arch/um/include/shared/kern_util.h      |  3 +-
 arch/um/include/shared/ptrace_user.h    | 41 ---------------
 arch/um/kernel/process.c                | 12 +----
 arch/um/kernel/ptrace.c                 |  2 -
 arch/um/kernel/signal.c                 | 12 -----
 arch/um/os-Linux/skas/process.c         | 60 ++++-----------------
 arch/um/os-Linux/start_up.c             | 70 +++----------------------
 arch/x86/um/ptrace_32.c                 | 24 ---------
 arch/x86/um/ptrace_64.c                 | 26 ---------
 arch/x86/um/shared/sysdep/ptrace_32.h   |  4 --
 arch/x86/um/shared/sysdep/ptrace_user.h | 12 ++---
 12 files changed, 24 insertions(+), 243 deletions(-)

diff --git a/arch/um/include/asm/processor-generic.h b/arch/um/include/asm/processor-generic.h
index 7414154b8e9a..6c3779541845 100644
--- a/arch/um/include/asm/processor-generic.h
+++ b/arch/um/include/asm/processor-generic.h
@@ -22,7 +22,6 @@ struct mm_struct;
 struct thread_struct {
 	struct pt_regs regs;
 	struct pt_regs *segv_regs;
-	int singlestep_syscall;
 	void *fault_addr;
 	jmp_buf *fault_catcher;
 	struct task_struct *prev_sched;
diff --git a/arch/um/include/shared/kern_util.h b/arch/um/include/shared/kern_util.h
index d8b8b4f07e42..11b4f3d7c18a 100644
--- a/arch/um/include/shared/kern_util.h
+++ b/arch/um/include/shared/kern_util.h
@@ -34,7 +34,6 @@ extern int handle_page_fault(unsigned long address, unsigned long ip,
 
 extern unsigned int do_IRQ(int irq, struct uml_pt_regs *regs);
 extern void initial_thread_cb(void (*proc)(void *), void *arg);
-extern int is_syscall(unsigned long addr);
 
 extern void timer_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs);
 
@@ -58,7 +57,7 @@ extern char *uml_strdup(const char *string);
 extern unsigned long to_irq_stack(unsigned long *mask_out);
 extern unsigned long from_irq_stack(int nested);
 
-extern int singlestepping(void *t);
+extern int singlestepping(void);
 
 extern void segv_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs);
 extern void bus_handler(int sig, struct siginfo *si, struct uml_pt_regs *regs);
diff --git a/arch/um/include/shared/ptrace_user.h b/arch/um/include/shared/ptrace_user.h
index 95455e8996e7..8a705d8f96ce 100644
--- a/arch/um/include/shared/ptrace_user.h
+++ b/arch/um/include/shared/ptrace_user.h
@@ -12,45 +12,4 @@
 extern int ptrace_getregs(long pid, unsigned long *regs_out);
 extern int ptrace_setregs(long pid, unsigned long *regs_in);
 
-/* syscall emulation path in ptrace */
-
-#ifndef PTRACE_SYSEMU
-#define PTRACE_SYSEMU 31
-#endif
-#ifndef PTRACE_SYSEMU_SINGLESTEP
-#define PTRACE_SYSEMU_SINGLESTEP 32
-#endif
-
-/* On architectures, that started to support PTRACE_O_TRACESYSGOOD
- * in linux 2.4, there are two different definitions of
- * PTRACE_SETOPTIONS: linux 2.4 uses 21 while linux 2.6 uses 0x4200.
- * For binary compatibility, 2.6 also supports the old "21", named
- * PTRACE_OLDSETOPTION. On these architectures, UML always must use
- * "21", to ensure the kernel runs on 2.4 and 2.6 host without
- * recompilation. So, we use PTRACE_OLDSETOPTIONS in UML.
- * We also want to be able to build the kernel on 2.4, which doesn't
- * have PTRACE_OLDSETOPTIONS. So, if it is missing, we declare
- * PTRACE_OLDSETOPTIONS to be the same as PTRACE_SETOPTIONS.
- *
- * On architectures, that start to support PTRACE_O_TRACESYSGOOD on
- * linux 2.6, PTRACE_OLDSETOPTIONS never is defined, and also isn't
- * supported by the host kernel. In that case, our trick lets us use
- * the new 0x4200 with the name PTRACE_OLDSETOPTIONS.
- */
-#ifndef PTRACE_OLDSETOPTIONS
-#define PTRACE_OLDSETOPTIONS PTRACE_SETOPTIONS
-#endif
-
-void set_using_sysemu(int value);
-int get_using_sysemu(void);
-extern int sysemu_supported;
-
-#define SELECT_PTRACE_OPERATION(sysemu_mode, singlestep_mode) \
-	(((int[3][3] ) { \
-		{ PTRACE_SYSCALL, PTRACE_SYSCALL, PTRACE_SINGLESTEP }, \
-		{ PTRACE_SYSEMU, PTRACE_SYSEMU, PTRACE_SINGLESTEP }, \
-		{ PTRACE_SYSEMU, PTRACE_SYSEMU_SINGLESTEP, \
-		  PTRACE_SYSEMU_SINGLESTEP } }) \
-		[sysemu_mode][singlestep_mode])
-
 #endif
diff --git a/arch/um/kernel/process.c b/arch/um/kernel/process.c
index 106b7da2f8d6..e10bcd6fb869 100644
--- a/arch/um/kernel/process.c
+++ b/arch/um/kernel/process.c
@@ -332,17 +332,9 @@ int __init make_proc_sysemu(void)
 
 late_initcall(make_proc_sysemu);
 
-int singlestepping(void * t)
+int singlestepping(void)
 {
-	struct task_struct *task = t ? t : current;
-
-	if (!test_thread_flag(TIF_SINGLESTEP))
-		return 0;
-
-	if (task->thread.singlestep_syscall)
-		return 1;
-
-	return 2;
+	return test_thread_flag(TIF_SINGLESTEP);
 }
 
 /*
diff --git a/arch/um/kernel/ptrace.c b/arch/um/kernel/ptrace.c
index 5154b27de580..6600a2782796 100644
--- a/arch/um/kernel/ptrace.c
+++ b/arch/um/kernel/ptrace.c
@@ -12,7 +12,6 @@
 void user_enable_single_step(struct task_struct *child)
 {
 	set_tsk_thread_flag(child, TIF_SINGLESTEP);
-	child->thread.singlestep_syscall = 0;
 
 #ifdef SUBARCH_SET_SINGLESTEPPING
 	SUBARCH_SET_SINGLESTEPPING(child, 1);
@@ -22,7 +21,6 @@ void user_enable_single_step(struct task_struct *child)
 void user_disable_single_step(struct task_struct *child)
 {
 	clear_tsk_thread_flag(child, TIF_SINGLESTEP);
-	child->thread.singlestep_syscall = 0;
 
 #ifdef SUBARCH_SET_SINGLESTEPPING
 	SUBARCH_SET_SINGLESTEPPING(child, 0);
diff --git a/arch/um/kernel/signal.c b/arch/um/kernel/signal.c
index ae4658f576ab..a56b44522766 100644
--- a/arch/um/kernel/signal.c
+++ b/arch/um/kernel/signal.c
@@ -120,18 +120,6 @@ void do_signal(struct pt_regs *regs)
 		}
 	}
 
-	/*
-	 * This closes a way to execute a system call on the host.  If
-	 * you set a breakpoint on a system call instruction and singlestep
-	 * from it, the tracing thread used to PTRACE_SINGLESTEP the process
-	 * rather than PTRACE_SYSCALL it, allowing the system call to execute
-	 * on the host.  The tracing thread will check this flag and
-	 * PTRACE_SYSCALL if necessary.
-	 */
-	if (test_thread_flag(TIF_SINGLESTEP))
-		current->thread.singlestep_syscall =
-			is_syscall(PT_REGS_IP(&current->thread.regs));
-
 	/*
 	 * if there's no signal to deliver, we just put the saved sigmask
 	 * back
diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c
index 9464833e741a..43168c51850c 100644
--- a/arch/um/os-Linux/skas/process.c
+++ b/arch/um/os-Linux/skas/process.c
@@ -177,48 +177,11 @@ static void handle_segv(int pid, struct uml_pt_regs *regs, unsigned long *aux_fp
 	segv(regs->faultinfo, 0, 1, NULL);
 }
 
-/*
- * To use the same value of using_sysemu as the caller, ask it that value
- * (in local_using_sysemu
- */
-static void handle_trap(int pid, struct uml_pt_regs *regs,
-			int local_using_sysemu)
+static void handle_trap(int pid, struct uml_pt_regs *regs)
 {
-	int err, status;
-
 	if ((UPT_IP(regs) >= STUB_START) && (UPT_IP(regs) < STUB_END))
 		fatal_sigsegv();
 
-	if (!local_using_sysemu)
-	{
-		err = ptrace(PTRACE_POKEUSER, pid, PT_SYSCALL_NR_OFFSET,
-			     __NR_getpid);
-		if (err < 0) {
-			printk(UM_KERN_ERR "%s - nullifying syscall failed, errno = %d\n",
-			       __func__, errno);
-			fatal_sigsegv();
-		}
-
-		err = ptrace(PTRACE_SYSCALL, pid, 0, 0);
-		if (err < 0) {
-			printk(UM_KERN_ERR "%s - continuing to end of syscall failed, errno = %d\n",
-			       __func__, errno);
-			fatal_sigsegv();
-		}
-
-		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED | __WALL));
-		if ((err < 0) || !WIFSTOPPED(status) ||
-		    (WSTOPSIG(status) != SIGTRAP + 0x80)) {
-			err = ptrace_dump_regs(pid);
-			if (err)
-				printk(UM_KERN_ERR "Failed to get registers from process, errno = %d\n",
-				       -err);
-			printk(UM_KERN_ERR "%s - failed to wait at end of syscall, errno = %d, status = %d\n",
-			       __func__, errno, status);
-			fatal_sigsegv();
-		}
-	}
-
 	handle_syscall(regs);
 }
 
@@ -355,10 +318,10 @@ int start_userspace(unsigned long stub_stack)
 		goto out_kill;
 	}
 
-	if (ptrace(PTRACE_OLDSETOPTIONS, pid, NULL,
+	if (ptrace(PTRACE_SETOPTIONS, pid, NULL,
 		   (void *) PTRACE_O_TRACESYSGOOD) < 0) {
 		err = -errno;
-		printk(UM_KERN_ERR "%s : PTRACE_OLDSETOPTIONS failed, errno = %d\n",
+		printk(UM_KERN_ERR "%s : PTRACE_SETOPTIONS failed, errno = %d\n",
 		       __func__, errno);
 		goto out_kill;
 	}
@@ -380,8 +343,6 @@ int start_userspace(unsigned long stub_stack)
 void userspace(struct uml_pt_regs *regs, unsigned long *aux_fp_regs)
 {
 	int err, status, op, pid = userspace_pid[0];
-	/* To prevent races if using_sysemu changes under us.*/
-	int local_using_sysemu;
 	siginfo_t si;
 
 	/* Handle any immediate reschedules or signals */
@@ -411,11 +372,10 @@ void userspace(struct uml_pt_regs *regs, unsigned long *aux_fp_regs)
 			fatal_sigsegv();
 		}
 
-		/* Now we set local_using_sysemu to be used for one loop */
-		local_using_sysemu = get_using_sysemu();
-
-		op = SELECT_PTRACE_OPERATION(local_using_sysemu,
-					     singlestepping(NULL));
+		if (singlestepping())
+			op = PTRACE_SYSEMU_SINGLESTEP;
+		else
+			op = PTRACE_SYSEMU;
 
 		if (ptrace(op, pid, 0, 0)) {
 			printk(UM_KERN_ERR "%s - ptrace continue failed, op = %d, errno = %d\n",
@@ -474,7 +434,7 @@ void userspace(struct uml_pt_regs *regs, unsigned long *aux_fp_regs)
 				else handle_segv(pid, regs, aux_fp_regs);
 				break;
 			case SIGTRAP + 0x80:
-			        handle_trap(pid, regs, local_using_sysemu);
+				handle_trap(pid, regs);
 				break;
 			case SIGTRAP:
 				relay_signal(SIGTRAP, (struct siginfo *)&si, regs);
@@ -597,10 +557,10 @@ int copy_context_skas0(unsigned long new_stack, int pid)
 		goto out_kill;
 	}
 
-	if (ptrace(PTRACE_OLDSETOPTIONS, pid, NULL,
+	if (ptrace(PTRACE_SETOPTIONS, pid, NULL,
 		   (void *)PTRACE_O_TRACESYSGOOD) < 0) {
 		err = -errno;
-		printk(UM_KERN_ERR "%s : PTRACE_OLDSETOPTIONS failed, errno = %d\n",
+		printk(UM_KERN_ERR "%s : PTRACE_SETOPTIONS failed, errno = %d\n",
 		       __func__, errno);
 		goto out_kill;
 	}
diff --git a/arch/um/os-Linux/start_up.c b/arch/um/os-Linux/start_up.c
index e3ee4db58b40..561c08e0cc9c 100644
--- a/arch/um/os-Linux/start_up.c
+++ b/arch/um/os-Linux/start_up.c
@@ -143,71 +143,16 @@ static int stop_ptraced_child(int pid, int exitcode, int mustexit)
 	return ret;
 }
 
-/* Changed only during early boot */
-static int force_sysemu_disabled = 0;
-
-static int __init nosysemu_cmd_param(char *str, int* add)
-{
-	force_sysemu_disabled = 1;
-	return 0;
-}
-
-__uml_setup("nosysemu", nosysemu_cmd_param,
-"nosysemu\n"
-"    Turns off syscall emulation patch for ptrace (SYSEMU).\n"
-"    SYSEMU is a performance-patch introduced by Laurent Vivier. It changes\n"
-"    behaviour of ptrace() and helps reduce host context switch rates.\n"
-"    To make it work, you need a kernel patch for your host, too.\n"
-"    See http://perso.wanadoo.fr/laurent.vivier/UML/ for further \n"
-"    information.\n\n");
-
 static void __init check_sysemu(void)
 {
-	unsigned long regs[MAX_REG_NR];
 	int pid, n, status, count=0;
 
-	os_info("Checking syscall emulation patch for ptrace...");
-	sysemu_supported = 0;
-	pid = start_ptraced_child();
-
-	if (ptrace(PTRACE_SYSEMU, pid, 0, 0) < 0)
-		goto fail;
-
-	CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
-	if (n < 0)
-		fatal_perror("check_sysemu : wait failed");
-	if (!WIFSTOPPED(status) || (WSTOPSIG(status) != SIGTRAP))
-		fatal("check_sysemu : expected SIGTRAP, got status = %d\n",
-		      status);
-
-	if (ptrace(PTRACE_GETREGS, pid, 0, regs) < 0)
-		fatal_perror("check_sysemu : PTRACE_GETREGS failed");
-	if (PT_SYSCALL_NR(regs) != __NR_getpid) {
-		non_fatal("check_sysemu got system call number %d, "
-			  "expected %d...", PT_SYSCALL_NR(regs), __NR_getpid);
-		goto fail;
-	}
-
-	n = ptrace(PTRACE_POKEUSER, pid, PT_SYSCALL_RET_OFFSET, os_getpid());
-	if (n < 0) {
-		non_fatal("check_sysemu : failed to modify system call "
-			  "return");
-		goto fail;
-	}
-
-	if (stop_ptraced_child(pid, 0, 0) < 0)
-		goto fail_stopped;
-
-	sysemu_supported = 1;
-	os_info("OK\n");
-	set_using_sysemu(!force_sysemu_disabled);
-
-	os_info("Checking advanced syscall emulation patch for ptrace...");
+	os_info("Checking syscall emulation for ptrace...");
 	pid = start_ptraced_child();
 
-	if ((ptrace(PTRACE_OLDSETOPTIONS, pid, 0,
+	if ((ptrace(PTRACE_SETOPTIONS, pid, 0,
 		   (void *) PTRACE_O_TRACESYSGOOD) < 0))
-		fatal_perror("check_sysemu: PTRACE_OLDSETOPTIONS failed");
+		fatal_perror("check_sysemu: PTRACE_SETOPTIONS failed");
 
 	while (1) {
 		count++;
@@ -243,17 +188,14 @@ static void __init check_sysemu(void)
 	if (stop_ptraced_child(pid, 0, 0) < 0)
 		goto fail_stopped;
 
-	sysemu_supported = 2;
 	os_info("OK\n");
 
-	if (!force_sysemu_disabled)
-		set_using_sysemu(sysemu_supported);
 	return;
 
 fail:
 	stop_ptraced_child(pid, 1, 0);
 fail_stopped:
-	non_fatal("missing\n");
+	fatal("missing\n");
 }
 
 static void __init check_ptrace(void)
@@ -263,9 +205,9 @@ static void __init check_ptrace(void)
 	os_info("Checking that ptrace can change system call numbers...");
 	pid = start_ptraced_child();
 
-	if ((ptrace(PTRACE_OLDSETOPTIONS, pid, 0,
+	if ((ptrace(PTRACE_SETOPTIONS, pid, 0,
 		   (void *) PTRACE_O_TRACESYSGOOD) < 0))
-		fatal_perror("check_ptrace: PTRACE_OLDSETOPTIONS failed");
+		fatal_perror("check_ptrace: PTRACE_SETOPTIONS failed");
 
 	while (1) {
 		if (ptrace(PTRACE_SYSCALL, pid, 0, 0) < 0)
diff --git a/arch/x86/um/ptrace_32.c b/arch/x86/um/ptrace_32.c
index 0bc4b73a9cde..7f1abde2c84b 100644
--- a/arch/x86/um/ptrace_32.c
+++ b/arch/x86/um/ptrace_32.c
@@ -25,30 +25,6 @@ void arch_switch_to(struct task_struct *to)
 		printk(KERN_WARNING "arch_switch_tls failed, errno = EINVAL\n");
 }
 
-int is_syscall(unsigned long addr)
-{
-	unsigned short instr;
-	int n;
-
-	n = copy_from_user(&instr, (void __user *) addr, sizeof(instr));
-	if (n) {
-		/* access_process_vm() grants access to vsyscall and stub,
-		 * while copy_from_user doesn't. Maybe access_process_vm is
-		 * slow, but that doesn't matter, since it will be called only
-		 * in case of singlestepping, if copy_from_user failed.
-		 */
-		n = access_process_vm(current, addr, &instr, sizeof(instr),
-				FOLL_FORCE);
-		if (n != sizeof(instr)) {
-			printk(KERN_ERR "is_syscall : failed to read "
-			       "instruction from 0x%lx\n", addr);
-			return 1;
-		}
-	}
-	/* int 0x80 or sysenter */
-	return (instr == 0x80cd) || (instr == 0x340f);
-}
-
 /* determines which flags the user has access to. */
 /* 1 = access 0 = no access */
 #define FLAG_MASK 0x00044dd5
diff --git a/arch/x86/um/ptrace_64.c b/arch/x86/um/ptrace_64.c
index 289d0159b041..aa68d83d3f44 100644
--- a/arch/x86/um/ptrace_64.c
+++ b/arch/x86/um/ptrace_64.c
@@ -188,32 +188,6 @@ int peek_user(struct task_struct *child, long addr, long data)
 	return put_user(tmp, (unsigned long *) data);
 }
 
-/* XXX Mostly copied from sys-i386 */
-int is_syscall(unsigned long addr)
-{
-	unsigned short instr;
-	int n;
-
-	n = copy_from_user(&instr, (void __user *) addr, sizeof(instr));
-	if (n) {
-		/*
-		 * access_process_vm() grants access to vsyscall and stub,
-		 * while copy_from_user doesn't. Maybe access_process_vm is
-		 * slow, but that doesn't matter, since it will be called only
-		 * in case of singlestepping, if copy_from_user failed.
-		 */
-		n = access_process_vm(current, addr, &instr, sizeof(instr),
-				FOLL_FORCE);
-		if (n != sizeof(instr)) {
-			printk("is_syscall : failed to read instruction from "
-			       "0x%lx\n", addr);
-			return 1;
-		}
-	}
-	/* sysenter */
-	return instr == 0x050f;
-}
-
 static int get_fpregs(struct user_i387_struct __user *buf, struct task_struct *child)
 {
 	int err, n, cpu = ((struct thread_info *) child->stack)->cpu;
diff --git a/arch/x86/um/shared/sysdep/ptrace_32.h b/arch/x86/um/shared/sysdep/ptrace_32.h
index db8478a83a09..0c4989842fbe 100644
--- a/arch/x86/um/shared/sysdep/ptrace_32.h
+++ b/arch/x86/um/shared/sysdep/ptrace_32.h
@@ -8,10 +8,6 @@
 
 #define MAX_FP_NR HOST_FPX_SIZE
 
-void set_using_sysemu(int value);
-int get_using_sysemu(void);
-extern int sysemu_supported;
-
 #define UPT_SYSCALL_ARG1(r) UPT_BX(r)
 #define UPT_SYSCALL_ARG2(r) UPT_CX(r)
 #define UPT_SYSCALL_ARG3(r) UPT_DX(r)
diff --git a/arch/x86/um/shared/sysdep/ptrace_user.h b/arch/x86/um/shared/sysdep/ptrace_user.h
index 44782bbad41e..1d1a824fa652 100644
--- a/arch/x86/um/shared/sysdep/ptrace_user.h
+++ b/arch/x86/um/shared/sysdep/ptrace_user.h
@@ -15,14 +15,12 @@
 #define FP_SIZE ((HOST_FPX_SIZE > HOST_FP_SIZE) ? HOST_FPX_SIZE : HOST_FP_SIZE)
 #else
 #define FP_SIZE HOST_FP_SIZE
+#endif
 
 /*
- * x86_64 FC3 doesn't define this in /usr/include/linux/ptrace.h even though
- * it's defined in the kernel's include/linux/ptrace.h. Additionally, use the
- * 2.4 name and value for 2.4 host compatibility.
+ * glibc before 2.27 does not include PTRACE_SYSEMU_SINGLESTEP in its enum,
+ * ensure we have a definition by (re-)defining it here.
  */
-#ifndef PTRACE_OLDSETOPTIONS
-#define PTRACE_OLDSETOPTIONS 21
-#endif
-
+#ifndef PTRACE_SYSEMU_SINGLESTEP
+#define PTRACE_SYSEMU_SINGLESTEP 32
 #endif
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 02/11] um: Drop NULL check from start_userspace

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

start_userspace is only called from exactly one location, and the passed
pointer for the userspace process stack cannot be NULL.

Remove the check, without changing the control flow.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/os-Linux/skas/process.c | 53 +++++++++++++++------------------
 1 file changed, 24 insertions(+), 29 deletions(-)

diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c
index 43168c51850c..ed20cefb2963 100644
--- a/arch/um/os-Linux/skas/process.c
+++ b/arch/um/os-Linux/skas/process.c
@@ -189,7 +189,7 @@ extern char __syscall_stub_start[];
 
 /**
  * userspace_tramp() - userspace trampoline
- * @stack:	pointer to the new userspace stack page, can be NULL, if? FIXME:
+ * @stack:	pointer to the new userspace stack page
  *
  * The userspace trampoline is used to setup a new userspace process in start_userspace() after it was clone()'ed.
  * This function will run on a temporary stack page.
@@ -204,9 +204,13 @@ extern char __syscall_stub_start[];
  */
 static int userspace_tramp(void *stack)
 {
+	struct sigaction sa;
 	void *addr;
 	int fd;
 	unsigned long long offset;
+	unsigned long segv_handler = STUB_CODE +
+				     (unsigned long) stub_segv_handler -
+				     (unsigned long) __syscall_stub_start;
 
 	ptrace(PTRACE_TRACEME, 0, 0, 0);
 
@@ -222,34 +226,25 @@ static int userspace_tramp(void *stack)
 		exit(1);
 	}
 
-	if (stack != NULL) {
-		fd = phys_mapping(uml_to_phys(stack), &offset);
-		addr = mmap((void *) STUB_DATA,
-			    STUB_DATA_PAGES * UM_KERN_PAGE_SIZE, PROT_READ | PROT_WRITE,
-			    MAP_FIXED | MAP_SHARED, fd, offset);
-		if (addr == MAP_FAILED) {
-			printk(UM_KERN_ERR "mapping segfault stack at 0x%lx failed, errno = %d\n",
-			       STUB_DATA, errno);
-			exit(1);
-		}
+	fd = phys_mapping(uml_to_phys(stack), &offset);
+	addr = mmap((void *) STUB_DATA,
+		    STUB_DATA_PAGES * UM_KERN_PAGE_SIZE, PROT_READ | PROT_WRITE,
+		    MAP_FIXED | MAP_SHARED, fd, offset);
+	if (addr == MAP_FAILED) {
+		printk(UM_KERN_ERR "mapping segfault stack at 0x%lx failed, errno = %d\n",
+		       STUB_DATA, errno);
+		exit(1);
 	}
-	if (stack != NULL) {
-		struct sigaction sa;
-
-		unsigned long v = STUB_CODE +
-				  (unsigned long) stub_segv_handler -
-				  (unsigned long) __syscall_stub_start;
-
-		set_sigstack((void *) STUB_DATA, STUB_DATA_PAGES * UM_KERN_PAGE_SIZE);
-		sigemptyset(&sa.sa_mask);
-		sa.sa_flags = SA_ONSTACK | SA_NODEFER | SA_SIGINFO;
-		sa.sa_sigaction = (void *) v;
-		sa.sa_restorer = NULL;
-		if (sigaction(SIGSEGV, &sa, NULL) < 0) {
-			printk(UM_KERN_ERR "%s - setting SIGSEGV handler failed - errno = %d\n",
-			       __func__, errno);
-			exit(1);
-		}
+
+	set_sigstack((void *) STUB_DATA, STUB_DATA_PAGES * UM_KERN_PAGE_SIZE);
+	sigemptyset(&sa.sa_mask);
+	sa.sa_flags = SA_ONSTACK | SA_NODEFER | SA_SIGINFO;
+	sa.sa_sigaction = (void *) segv_handler;
+	sa.sa_restorer = NULL;
+	if (sigaction(SIGSEGV, &sa, NULL) < 0) {
+		printk(UM_KERN_ERR "%s - setting SIGSEGV handler failed - errno = %d\n",
+		       __func__, errno);
+		exit(1);
 	}
 
 	kill(os_getpid(), SIGSTOP);
@@ -261,7 +256,7 @@ int kill_userspace_mm[NR_CPUS];
 
 /**
  * start_userspace() - prepare a new userspace process
- * @stub_stack:	pointer to the stub stack. Can be NULL, if? FIXME:
+ * @stub_stack:	pointer to the stub stack.
  *
  * Setups a new temporary stack page that is used while userspace_tramp() runs
  * Clones the kernel process into a new userspace process, with FDs only.
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 03/11] um: Make errors to stop ptraced child fatal during startup

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

For the detection code to check whether SYSEMU_SINGLESTEP works
correctly we needed some error cases while stopping to be non-fatal.
However, at this point stop_ptraced_child must always succeed, and we
can therefore simplify it slightly to exit immediately on error.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/os-Linux/start_up.c | 41 +++++++++++--------------------------
 1 file changed, 12 insertions(+), 29 deletions(-)

diff --git a/arch/um/os-Linux/start_up.c b/arch/um/os-Linux/start_up.c
index 561c08e0cc9c..8b0e98ab842c 100644
--- a/arch/um/os-Linux/start_up.c
+++ b/arch/um/os-Linux/start_up.c
@@ -112,35 +112,20 @@ static int start_ptraced_child(void)
 	return pid;
 }
 
-/* When testing for SYSEMU support, if it is one of the broken versions, we
- * must just avoid using sysemu, not panic, but only if SYSEMU features are
- * broken.
- * So only for SYSEMU features we test mustpanic, while normal host features
- * must work anyway!
- */
-static int stop_ptraced_child(int pid, int exitcode, int mustexit)
+static void stop_ptraced_child(int pid, int exitcode)
 {
-	int status, n, ret = 0;
+	int status, n;
+
+	if (ptrace(PTRACE_CONT, pid, 0, 0) < 0)
+		fatal_perror("stop_ptraced_child : ptrace failed");
 
-	if (ptrace(PTRACE_CONT, pid, 0, 0) < 0) {
-		perror("stop_ptraced_child : ptrace failed");
-		return -1;
-	}
 	CATCH_EINTR(n = waitpid(pid, &status, 0));
 	if (!WIFEXITED(status) || (WEXITSTATUS(status) != exitcode)) {
 		int exit_with = WEXITSTATUS(status);
-		if (exit_with == 2)
-			non_fatal("check_ptrace : child exited with status 2. "
-				  "\nDisabling SYSEMU support.\n");
-		non_fatal("check_ptrace : child exited with exitcode %d, while "
-			  "expecting %d; status 0x%x\n", exit_with,
-			  exitcode, status);
-		if (mustexit)
-			exit(1);
-		ret = -1;
+		fatal("stop_ptraced_child : child exited with exitcode %d, "
+		      "while expecting %d; status 0x%x\n", exit_with,
+		      exitcode, status);
 	}
-
-	return ret;
 }
 
 static void __init check_sysemu(void)
@@ -185,16 +170,14 @@ static void __init check_sysemu(void)
 			goto fail;
 		}
 	}
-	if (stop_ptraced_child(pid, 0, 0) < 0)
-		goto fail_stopped;
+	stop_ptraced_child(pid, 0);
 
 	os_info("OK\n");
 
 	return;
 
 fail:
-	stop_ptraced_child(pid, 1, 0);
-fail_stopped:
+	stop_ptraced_child(pid, 1);
 	fatal("missing\n");
 }
 
@@ -233,7 +216,7 @@ static void __init check_ptrace(void)
 			break;
 		}
 	}
-	stop_ptraced_child(pid, 0, 1);
+	stop_ptraced_child(pid, 0);
 	os_info("OK\n");
 	check_sysemu();
 }
@@ -312,7 +295,7 @@ void __init os_early_checks(void)
 	pid = start_ptraced_child();
 	if (init_pid_registers(pid))
 		fatal("Failed to initialize default registers");
-	stop_ptraced_child(pid, 1, 1);
+	stop_ptraced_child(pid, 1);
 }
 
 int __init parse_iomem(char *str, int *add)
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 04/11] um: Don't use vfprintf() for os_info()

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

The threads allocated inside the kernel have only a single page of
stack. Unfortunately, the vfprintf function in standard glibc may use
too much stack-space, overflowing it.

To make os_info safe to be used by helper threads, use the kernel
vscnprintf function into a smallish buffer and write out the information
to stderr.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/os-Linux/util.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/arch/um/os-Linux/util.c b/arch/um/os-Linux/util.c
index fc0f2a9dee5a..1dca4ffbd572 100644
--- a/arch/um/os-Linux/util.c
+++ b/arch/um/os-Linux/util.c
@@ -173,23 +173,38 @@ __uml_setup("quiet", quiet_cmd_param,
 "quiet\n"
 "    Turns off information messages during boot.\n\n");
 
+/*
+ * The os_info/os_warn functions will be called by helper threads. These
+ * have a very limited stack size and using the libc formatting functions
+ * may overflow the stack.
+ * So pull in the kernel vscnprintf and use that instead with a fixed
+ * on-stack buffer.
+ */
+int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
+
 void os_info(const char *fmt, ...)
 {
+	char buf[256];
 	va_list list;
+	int len;
 
 	if (quiet_info)
 		return;
 
 	va_start(list, fmt);
-	vfprintf(stderr, fmt, list);
+	len = vscnprintf(buf, sizeof(buf), fmt, list);
+	fwrite(buf, len, 1, stderr);
 	va_end(list);
 }
 
 void os_warn(const char *fmt, ...)
 {
+	char buf[256];
 	va_list list;
+	int len;
 
 	va_start(list, fmt);
-	vfprintf(stderr, fmt, list);
+	len = vscnprintf(buf, sizeof(buf), fmt, list);
+	fwrite(buf, len, 1, stderr);
 	va_end(list);
 }
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 05/11] um: Do not use printk in SIGWINCH helper thread

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

The thread is running outside of the UML kernel scope and is a helper.
As such, printk cannot work and os_info must be used instead.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/drivers/chan_user.c | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/arch/um/drivers/chan_user.c b/arch/um/drivers/chan_user.c
index 25727ed648b7..c2b83cb99aae 100644
--- a/arch/um/drivers/chan_user.c
+++ b/arch/um/drivers/chan_user.c
@@ -153,8 +153,8 @@ static int winch_thread(void *arg)
 	pipe_fd = data->pipe_fd;
 	count = write(pipe_fd, &c, sizeof(c));
 	if (count != sizeof(c))
-		printk(UM_KERN_ERR "winch_thread : failed to write "
-		       "synchronization byte, err = %d\n", -count);
+		os_info("winch_thread : failed to write synchronization byte, err = %d\n",
+			-count);
 
 	/*
 	 * We are not using SIG_IGN on purpose, so don't fix it as I thought to
@@ -166,28 +166,28 @@ static int winch_thread(void *arg)
 	sigfillset(&sigs);
 	/* Block all signals possible. */
 	if (sigprocmask(SIG_SETMASK, &sigs, NULL) < 0) {
-		printk(UM_KERN_ERR "winch_thread : sigprocmask failed, "
-		       "errno = %d\n", errno);
+		os_info("winch_thread : sigprocmask failed, errno = %d\n",
+			errno);
 		exit(1);
 	}
 	/* In sigsuspend(), block anything else than SIGWINCH. */
 	sigdelset(&sigs, SIGWINCH);
 
 	if (setsid() < 0) {
-		printk(UM_KERN_ERR "winch_thread : setsid failed, errno = %d\n",
+		os_info("winch_thread : setsid failed, errno = %d\n",
 		       errno);
 		exit(1);
 	}
 
 	if (ioctl(pty_fd, TIOCSCTTY, 0) < 0) {
-		printk(UM_KERN_ERR "winch_thread : TIOCSCTTY failed on "
-		       "fd %d err = %d\n", pty_fd, errno);
+		os_info("winch_thread : TIOCSCTTY failed on "
+			"fd %d err = %d\n", pty_fd, errno);
 		exit(1);
 	}
 
 	if (tcsetpgrp(pty_fd, os_getpid()) < 0) {
-		printk(UM_KERN_ERR "winch_thread : tcsetpgrp failed on "
-		       "fd %d err = %d\n", pty_fd, errno);
+		os_info("winch_thread : tcsetpgrp failed on fd %d err = %d\n",
+			pty_fd, errno);
 		exit(1);
 	}
 
@@ -199,8 +199,8 @@ static int winch_thread(void *arg)
 	 */
 	count = read(pipe_fd, &c, sizeof(c));
 	if (count != sizeof(c))
-		printk(UM_KERN_ERR "winch_thread : failed to read "
-		       "synchronization byte, err = %d\n", errno);
+		os_info("winch_thread : failed to read synchronization byte, err = %d\n",
+			errno);
 
 	while(1) {
 		/*
@@ -211,8 +211,8 @@ static int winch_thread(void *arg)
 
 		count = write(pipe_fd, &c, sizeof(c));
 		if (count != sizeof(c))
-			printk(UM_KERN_ERR "winch_thread : write failed, "
-			       "err = %d\n", errno);
+			os_info("winch_thread : write failed, err = %d\n",
+				errno);
 	}
 }
 
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 06/11] um: Reap winch thread if it fails

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

When the winch thread runs into an error condition, it would exit(1) and
never be reaped until shutdown time. Change this to write a command byte
which causes the driver to kill it, therefore reaping the child.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/drivers/chan_user.c | 16 +++++++++++-----
 arch/um/drivers/line.c      | 13 ++++++++-----
 2 files changed, 19 insertions(+), 10 deletions(-)

diff --git a/arch/um/drivers/chan_user.c b/arch/um/drivers/chan_user.c
index c2b83cb99aae..ed7cc830b3e7 100644
--- a/arch/um/drivers/chan_user.c
+++ b/arch/um/drivers/chan_user.c
@@ -141,7 +141,7 @@ struct winch_data {
 	int pipe_fd;
 };
 
-static int winch_thread(void *arg)
+static __noreturn int winch_thread(void *arg)
 {
 	struct winch_data *data = arg;
 	sigset_t sigs;
@@ -168,7 +168,7 @@ static int winch_thread(void *arg)
 	if (sigprocmask(SIG_SETMASK, &sigs, NULL) < 0) {
 		os_info("winch_thread : sigprocmask failed, errno = %d\n",
 			errno);
-		exit(1);
+		goto wait_kill;
 	}
 	/* In sigsuspend(), block anything else than SIGWINCH. */
 	sigdelset(&sigs, SIGWINCH);
@@ -176,19 +176,19 @@ static int winch_thread(void *arg)
 	if (setsid() < 0) {
 		os_info("winch_thread : setsid failed, errno = %d\n",
 		       errno);
-		exit(1);
+		goto wait_kill;
 	}
 
 	if (ioctl(pty_fd, TIOCSCTTY, 0) < 0) {
 		os_info("winch_thread : TIOCSCTTY failed on "
 			"fd %d err = %d\n", pty_fd, errno);
-		exit(1);
+		goto wait_kill;
 	}
 
 	if (tcsetpgrp(pty_fd, os_getpid()) < 0) {
 		os_info("winch_thread : tcsetpgrp failed on fd %d err = %d\n",
 			pty_fd, errno);
-		exit(1);
+		goto wait_kill;
 	}
 
 	/*
@@ -214,6 +214,12 @@ static int winch_thread(void *arg)
 			os_info("winch_thread : write failed, err = %d\n",
 				errno);
 	}
+
+wait_kill:
+	c = 2;
+	count = write(pipe_fd, &c, sizeof(c));
+	while (1)
+		pause();
 }
 
 static int winch_tramp(int fd, struct tty_port *port, int *fd_out,
diff --git a/arch/um/drivers/line.c b/arch/um/drivers/line.c
index b98545f3edb5..449d320c3f55 100644
--- a/arch/um/drivers/line.c
+++ b/arch/um/drivers/line.c
@@ -629,15 +629,18 @@ static irqreturn_t winch_interrupt(int irq, void *data)
 
 	if (fd != -1) {
 		err = generic_read(fd, &c, NULL);
-		if (err < 0) {
+		/* A read of 2 means the winch thread failed and has warned */
+		if (err < 0 || (err == 1 && c == 2)) {
 			if (err != -EAGAIN) {
 				winch->fd = -1;
 				list_del(&winch->list);
 				os_close_file(fd);
-				printk(KERN_ERR "winch_interrupt : "
-				       "read failed, errno = %d\n", -err);
-				printk(KERN_ERR "fd %d is losing SIGWINCH "
-				       "support\n", winch->tty_fd);
+				if (err < 0) {
+					printk(KERN_ERR "winch_interrupt : read failed, errno = %d\n",
+					       -err);
+					printk(KERN_ERR "fd %d is losing SIGWINCH support\n",
+					       winch->tty_fd);
+				}
 				INIT_WORK(&winch->work, __free_winch);
 				schedule_work(&winch->work);
 				return IRQ_HANDLED;
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 07/11] um: Do not use printk in userspace trampoline

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

The trampoline is running in a cloned process. It is not safe to use
printk for error printing there.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/os-Linux/skas/process.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c
index ed20cefb2963..1f5c3f2523d1 100644
--- a/arch/um/os-Linux/skas/process.c
+++ b/arch/um/os-Linux/skas/process.c
@@ -221,8 +221,8 @@ static int userspace_tramp(void *stack)
 	addr = mmap64((void *) STUB_CODE, UM_KERN_PAGE_SIZE,
 		      PROT_EXEC, MAP_FIXED | MAP_PRIVATE, fd, offset);
 	if (addr == MAP_FAILED) {
-		printk(UM_KERN_ERR "mapping mmap stub at 0x%lx failed, errno = %d\n",
-		       STUB_CODE, errno);
+		os_info("mapping mmap stub at 0x%lx failed, errno = %d\n",
+			STUB_CODE, errno);
 		exit(1);
 	}
 
@@ -231,8 +231,8 @@ static int userspace_tramp(void *stack)
 		    STUB_DATA_PAGES * UM_KERN_PAGE_SIZE, PROT_READ | PROT_WRITE,
 		    MAP_FIXED | MAP_SHARED, fd, offset);
 	if (addr == MAP_FAILED) {
-		printk(UM_KERN_ERR "mapping segfault stack at 0x%lx failed, errno = %d\n",
-		       STUB_DATA, errno);
+		os_info("mapping segfault stack at 0x%lx failed, errno = %d\n",
+			STUB_DATA, errno);
 		exit(1);
 	}
 
@@ -242,8 +242,8 @@ static int userspace_tramp(void *stack)
 	sa.sa_sigaction = (void *) segv_handler;
 	sa.sa_restorer = NULL;
 	if (sigaction(SIGSEGV, &sa, NULL) < 0) {
-		printk(UM_KERN_ERR "%s - setting SIGSEGV handler failed - errno = %d\n",
-		       __func__, errno);
+		os_info("%s - setting SIGSEGV handler failed - errno = %d\n",
+			__func__, errno);
 		exit(1);
 	}
 
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 08/11] um: Always inline stub functions

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

The stub executable page is remapped to a different location in the
userland process. As these functions may be used by the stub, they
really need to be always inlined rather than permitting the compiler to
emit a function.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/x86/um/shared/sysdep/stub_32.h | 21 +++++++++++----------
 arch/x86/um/shared/sysdep/stub_64.h | 17 +++++++++--------
 2 files changed, 20 insertions(+), 18 deletions(-)

diff --git a/arch/x86/um/shared/sysdep/stub_32.h b/arch/x86/um/shared/sysdep/stub_32.h
index 38fa894b65d0..4e763f8b380f 100644
--- a/arch/x86/um/shared/sysdep/stub_32.h
+++ b/arch/x86/um/shared/sysdep/stub_32.h
@@ -12,7 +12,7 @@
 #define STUB_MMAP_NR __NR_mmap2
 #define MMAP_OFFSET(o) ((o) >> UM_KERN_PAGE_SHIFT)
 
-static inline long stub_syscall0(long syscall)
+static __always_inline long stub_syscall0(long syscall)
 {
 	long ret;
 
@@ -21,7 +21,7 @@ static inline long stub_syscall0(long syscall)
 	return ret;
 }
 
-static inline long stub_syscall1(long syscall, long arg1)
+static __always_inline long stub_syscall1(long syscall, long arg1)
 {
 	long ret;
 
@@ -30,7 +30,7 @@ static inline long stub_syscall1(long syscall, long arg1)
 	return ret;
 }
 
-static inline long stub_syscall2(long syscall, long arg1, long arg2)
+static __always_inline long stub_syscall2(long syscall, long arg1, long arg2)
 {
 	long ret;
 
@@ -40,7 +40,8 @@ static inline long stub_syscall2(long syscall, long arg1, long arg2)
 	return ret;
 }
 
-static inline long stub_syscall3(long syscall, long arg1, long arg2, long arg3)
+static __always_inline long stub_syscall3(long syscall, long arg1, long arg2,
+					  long arg3)
 {
 	long ret;
 
@@ -50,8 +51,8 @@ static inline long stub_syscall3(long syscall, long arg1, long arg2, long arg3)
 	return ret;
 }
 
-static inline long stub_syscall4(long syscall, long arg1, long arg2, long arg3,
-				 long arg4)
+static __always_inline long stub_syscall4(long syscall, long arg1, long arg2,
+					  long arg3, long arg4)
 {
 	long ret;
 
@@ -61,8 +62,8 @@ static inline long stub_syscall4(long syscall, long arg1, long arg2, long arg3,
 	return ret;
 }
 
-static inline long stub_syscall5(long syscall, long arg1, long arg2, long arg3,
-				 long arg4, long arg5)
+static __always_inline long stub_syscall5(long syscall, long arg1, long arg2,
+					  long arg3, long arg4, long arg5)
 {
 	long ret;
 
@@ -72,12 +73,12 @@ static inline long stub_syscall5(long syscall, long arg1, long arg2, long arg3,
 	return ret;
 }
 
-static inline void trap_myself(void)
+static __always_inline void trap_myself(void)
 {
 	__asm("int3");
 }
 
-static inline void remap_stack_and_trap(void)
+static __always_inline void remap_stack_and_trap(void)
 {
 	__asm__ volatile (
 		"movl %%esp,%%ebx ;"
diff --git a/arch/x86/um/shared/sysdep/stub_64.h b/arch/x86/um/shared/sysdep/stub_64.h
index 2de1c8f88173..b24168ef0ac4 100644
--- a/arch/x86/um/shared/sysdep/stub_64.h
+++ b/arch/x86/um/shared/sysdep/stub_64.h
@@ -16,7 +16,7 @@
 #define __syscall_clobber "r11","rcx","memory"
 #define __syscall "syscall"
 
-static inline long stub_syscall0(long syscall)
+static __always_inline long stub_syscall0(long syscall)
 {
 	long ret;
 
@@ -27,7 +27,7 @@ static inline long stub_syscall0(long syscall)
 	return ret;
 }
 
-static inline long stub_syscall2(long syscall, long arg1, long arg2)
+static __always_inline long stub_syscall2(long syscall, long arg1, long arg2)
 {
 	long ret;
 
@@ -38,7 +38,8 @@ static inline long stub_syscall2(long syscall, long arg1, long arg2)
 	return ret;
 }
 
-static inline long stub_syscall3(long syscall, long arg1, long arg2, long arg3)
+static __always_inline long stub_syscall3(long syscall, long arg1, long arg2,
+					  long arg3)
 {
 	long ret;
 
@@ -50,7 +51,7 @@ static inline long stub_syscall3(long syscall, long arg1, long arg2, long arg3)
 	return ret;
 }
 
-static inline long stub_syscall4(long syscall, long arg1, long arg2, long arg3,
+static __always_inline long stub_syscall4(long syscall, long arg1, long arg2, long arg3,
 				 long arg4)
 {
 	long ret;
@@ -64,8 +65,8 @@ static inline long stub_syscall4(long syscall, long arg1, long arg2, long arg3,
 	return ret;
 }
 
-static inline long stub_syscall5(long syscall, long arg1, long arg2, long arg3,
-				 long arg4, long arg5)
+static __always_inline long stub_syscall5(long syscall, long arg1, long arg2,
+					  long arg3, long arg4, long arg5)
 {
 	long ret;
 
@@ -78,12 +79,12 @@ static inline long stub_syscall5(long syscall, long arg1, long arg2, long arg3,
 	return ret;
 }
 
-static inline void trap_myself(void)
+static __always_inline void trap_myself(void)
 {
 	__asm("int3");
 }
 
-static inline void remap_stack_and_trap(void)
+static __always_inline void remap_stack_and_trap(void)
 {
 	__asm__ volatile (
 		"movq %0,%%rax ;"
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 09/11] um: Rely on PTRACE_SETREGSET to set FS/GS base registers

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

These registers are saved/restored together with the other general
registers using ptrace. In arch_set_tls we then just need to set the
register and it will be synced back normally.

Most of this logic was introduced in commit f355559cf7845 ("[PATCH] uml:
x86_64 thread fixes"). However, at least today we can rely on ptrace to
restore the base registers for us. As such, only the part of the patch
that tracks the FS register for use as thread local storage is actually
needed.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/include/shared/os.h    |  3 --
 arch/x86/um/asm/elf.h          |  4 +--
 arch/x86/um/asm/processor_64.h |  3 --
 arch/x86/um/os-Linux/Makefile  |  1 -
 arch/x86/um/os-Linux/prctl.c   | 12 -------
 arch/x86/um/syscalls_64.c      | 62 +++++++---------------------------
 arch/x86/um/tls_64.c           |  2 +-
 7 files changed, 16 insertions(+), 71 deletions(-)
 delete mode 100644 arch/x86/um/os-Linux/prctl.c

diff --git a/arch/um/include/shared/os.h b/arch/um/include/shared/os.h
index 0df646c6651e..aff8906304ea 100644
--- a/arch/um/include/shared/os.h
+++ b/arch/um/include/shared/os.h
@@ -323,9 +323,6 @@ extern void sigio_broken(int fd);
 extern int __add_sigio_fd(int fd);
 extern int __ignore_sigio_fd(int fd);
 
-/* prctl.c */
-extern int os_arch_prctl(int pid, int option, unsigned long *arg2);
-
 /* tty.c */
 extern int get_pty(void);
 
diff --git a/arch/x86/um/asm/elf.h b/arch/x86/um/asm/elf.h
index 6523eb7c3bd1..6052200fe925 100644
--- a/arch/x86/um/asm/elf.h
+++ b/arch/x86/um/asm/elf.h
@@ -168,8 +168,8 @@ do {								\
 	(pr_reg)[18] = (_regs)->regs.gp[18];			\
 	(pr_reg)[19] = (_regs)->regs.gp[19];			\
 	(pr_reg)[20] = (_regs)->regs.gp[20];			\
-	(pr_reg)[21] = current->thread.arch.fs;			\
-	(pr_reg)[22] = 0;					\
+	(pr_reg)[21] = (_regs)->regs.gp[21];			\
+	(pr_reg)[22] = (_regs)->regs.gp[22];			\
 	(pr_reg)[23] = 0;					\
 	(pr_reg)[24] = 0;					\
 	(pr_reg)[25] = 0;					\
diff --git a/arch/x86/um/asm/processor_64.h b/arch/x86/um/asm/processor_64.h
index 1ef9c21877bc..f90159508936 100644
--- a/arch/x86/um/asm/processor_64.h
+++ b/arch/x86/um/asm/processor_64.h
@@ -10,13 +10,11 @@
 struct arch_thread {
         unsigned long debugregs[8];
         int debugregs_seq;
-        unsigned long fs;
         struct faultinfo faultinfo;
 };
 
 #define INIT_ARCH_THREAD { .debugregs  		= { [ 0 ... 7 ] = 0 }, \
 			   .debugregs_seq	= 0, \
-			   .fs			= 0, \
 			   .faultinfo		= { 0, 0, 0 } }
 
 #define STACKSLOTS_PER_LINE 4
@@ -28,7 +26,6 @@ static inline void arch_flush_thread(struct arch_thread *thread)
 static inline void arch_copy_thread(struct arch_thread *from,
                                     struct arch_thread *to)
 {
-	to->fs = from->fs;
 }
 
 #define current_sp() ({ void *sp; __asm__("movq %%rsp, %0" : "=r" (sp) : ); sp; })
diff --git a/arch/x86/um/os-Linux/Makefile b/arch/x86/um/os-Linux/Makefile
index ae169125d03f..5249bbc30dcd 100644
--- a/arch/x86/um/os-Linux/Makefile
+++ b/arch/x86/um/os-Linux/Makefile
@@ -6,7 +6,6 @@
 obj-y = registers.o task_size.o mcontext.o
 
 obj-$(CONFIG_X86_32) += tls.o
-obj-$(CONFIG_64BIT) += prctl.o
 
 USER_OBJS := $(obj-y)
 
diff --git a/arch/x86/um/os-Linux/prctl.c b/arch/x86/um/os-Linux/prctl.c
deleted file mode 100644
index 8431e87ac333..000000000000
--- a/arch/x86/um/os-Linux/prctl.c
+++ /dev/null
@@ -1,12 +0,0 @@
-/*
- * Copyright (C) 2007 Jeff Dike (jdike@{addtoit.com,linux.intel.com})
- * Licensed under the GPL
- */
-
-#include <sys/ptrace.h>
-#include <asm/ptrace.h>
-
-int os_arch_prctl(int pid, int option, unsigned long *arg2)
-{
-	return ptrace(PTRACE_ARCH_PRCTL, pid, (unsigned long) arg2, option);
-}
diff --git a/arch/x86/um/syscalls_64.c b/arch/x86/um/syscalls_64.c
index 27b29ae6c471..6a00a28c9cca 100644
--- a/arch/x86/um/syscalls_64.c
+++ b/arch/x86/um/syscalls_64.c
@@ -16,60 +16,24 @@
 long arch_prctl(struct task_struct *task, int option,
 		unsigned long __user *arg2)
 {
-	unsigned long *ptr = arg2, tmp;
-	long ret;
-	int pid = task->mm->context.id.u.pid;
-
-	/*
-	 * With ARCH_SET_FS (and ARCH_SET_GS is treated similarly to
-	 * be safe), we need to call arch_prctl on the host because
-	 * setting %fs may result in something else happening (like a
-	 * GDT or thread.fs being set instead).  So, we let the host
-	 * fiddle the registers and thread struct and restore the
-	 * registers afterwards.
-	 *
-	 * So, the saved registers are stored to the process (this
-	 * needed because a stub may have been the last thing to run),
-	 * arch_prctl is run on the host, then the registers are read
-	 * back.
-	 */
-	switch (option) {
-	case ARCH_SET_FS:
-	case ARCH_SET_GS:
-		ret = restore_pid_registers(pid, &current->thread.regs.regs);
-		if (ret)
-			return ret;
-		break;
-	case ARCH_GET_FS:
-	case ARCH_GET_GS:
-		/*
-		 * With these two, we read to a local pointer and
-		 * put_user it to the userspace pointer that we were
-		 * given.  If addr isn't valid (because it hasn't been
-		 * faulted in or is just bogus), we want put_user to
-		 * fault it in (or return -EFAULT) instead of having
-		 * the host return -EFAULT.
-		 */
-		ptr = &tmp;
-	}
-
-	ret = os_arch_prctl(pid, option, ptr);
-	if (ret)
-		return ret;
+	long ret = -EINVAL;
 
 	switch (option) {
 	case ARCH_SET_FS:
-		current->thread.arch.fs = (unsigned long) ptr;
-		ret = save_registers(pid, &current->thread.regs.regs);
+		current->thread.regs.regs.gp[FS_BASE / sizeof(unsigned long)] =
+			(unsigned long) arg2;
+		ret = 0;
 		break;
 	case ARCH_SET_GS:
-		ret = save_registers(pid, &current->thread.regs.regs);
+		current->thread.regs.regs.gp[GS_BASE / sizeof(unsigned long)] =
+			(unsigned long) arg2;
+		ret = 0;
 		break;
 	case ARCH_GET_FS:
-		ret = put_user(tmp, arg2);
+		ret = put_user(current->thread.regs.regs.gp[FS_BASE / sizeof(unsigned long)], arg2);
 		break;
 	case ARCH_GET_GS:
-		ret = put_user(tmp, arg2);
+		ret = put_user(current->thread.regs.regs.gp[GS_BASE / sizeof(unsigned long)], arg2);
 		break;
 	}
 
@@ -83,10 +47,10 @@ SYSCALL_DEFINE2(arch_prctl, int, option, unsigned long, arg2)
 
 void arch_switch_to(struct task_struct *to)
 {
-	if ((to->thread.arch.fs == 0) || (to->mm == NULL))
-		return;
-
-	arch_prctl(to, ARCH_SET_FS, (void __user *) to->thread.arch.fs);
+	/*
+	 * Nothing needs to be done on x86_64.
+	 * The FS_BASE/GS_BASE registers are saved in the ptrace register set.
+	 */
 }
 
 SYSCALL_DEFINE6(mmap, unsigned long, addr, unsigned long, len,
diff --git a/arch/x86/um/tls_64.c b/arch/x86/um/tls_64.c
index ebd3855d9b13..c51a613f6f5c 100644
--- a/arch/x86/um/tls_64.c
+++ b/arch/x86/um/tls_64.c
@@ -12,7 +12,7 @@ int arch_set_tls(struct task_struct *t, unsigned long tls)
 	 * If CLONE_SETTLS is set, we need to save the thread id
 	 * so it can be set during context switches.
 	 */
-	t->thread.arch.fs = tls;
+	t->thread.regs.regs.gp[FS_BASE / sizeof(unsigned long)] = tls;
 
 	return 0;
 }
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 10/11] um: Remove unused register save/restore functions

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

These functions were only used when calling PTRACE_ARCH_PRCTL, but this
code has been removed.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/um/include/shared/registers.h |  2 --
 arch/um/os-Linux/registers.c       | 20 --------------------
 2 files changed, 22 deletions(-)

diff --git a/arch/um/include/shared/registers.h b/arch/um/include/shared/registers.h
index 2f9c3ce5b45e..a0450326521c 100644
--- a/arch/um/include/shared/registers.h
+++ b/arch/um/include/shared/registers.h
@@ -14,8 +14,6 @@ extern int save_fp_registers(int pid, unsigned long *fp_regs);
 extern int restore_fp_registers(int pid, unsigned long *fp_regs);
 extern int save_fpx_registers(int pid, unsigned long *fp_regs);
 extern int restore_fpx_registers(int pid, unsigned long *fp_regs);
-extern int save_registers(int pid, struct uml_pt_regs *regs);
-extern int restore_pid_registers(int pid, struct uml_pt_regs *regs);
 extern int init_pid_registers(int pid);
 extern void get_safe_registers(unsigned long *regs, unsigned long *fp_regs);
 extern int get_fp_registers(int pid, unsigned long *regs);
diff --git a/arch/um/os-Linux/registers.c b/arch/um/os-Linux/registers.c
index b123955be7ac..bd80b921add0 100644
--- a/arch/um/os-Linux/registers.c
+++ b/arch/um/os-Linux/registers.c
@@ -11,26 +11,6 @@
 #include <sysdep/ptrace_user.h>
 #include <registers.h>
 
-int save_registers(int pid, struct uml_pt_regs *regs)
-{
-	int err;
-
-	err = ptrace(PTRACE_GETREGS, pid, 0, regs->gp);
-	if (err < 0)
-		return -errno;
-	return 0;
-}
-
-int restore_pid_registers(int pid, struct uml_pt_regs *regs)
-{
-	int err;
-
-	err = ptrace(PTRACE_SETREGS, pid, 0, regs->gp);
-	if (err < 0)
-		return -errno;
-	return 0;
-}
-
 /* This is set once at boot time and not changed thereafter */
 
 static unsigned long exec_regs[MAX_REG_NR];
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

[PATCH v3 11/11] um: Mark 32bit syscall helpers as clobbering memory

[benjamin]

From: Benjamin Berg <benjamin@sipsolutions.net>

The 64bit helper are marked to clobber the memory, but the 32bit ones
are not. Add the appropriate clobber to the 32bit helper routines so
that the compiler cannot do invalid optimizations.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
---
 arch/x86/um/shared/sysdep/stub_32.h | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/arch/x86/um/shared/sysdep/stub_32.h b/arch/x86/um/shared/sysdep/stub_32.h
index 4e763f8b380f..ea8b5a2d67af 100644
--- a/arch/x86/um/shared/sysdep/stub_32.h
+++ b/arch/x86/um/shared/sysdep/stub_32.h
@@ -16,7 +16,8 @@ static __always_inline long stub_syscall0(long syscall)
 {
 	long ret;
 
-	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall));
+	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall)
+			: "memory");
 
 	return ret;
 }
@@ -25,7 +26,8 @@ static __always_inline long stub_syscall1(long syscall, long arg1)
 {
 	long ret;
 
-	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1));
+	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1)
+			: "memory");
 
 	return ret;
 }
@@ -35,7 +37,8 @@ static __always_inline long stub_syscall2(long syscall, long arg1, long arg2)
 	long ret;
 
 	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1),
-			"c" (arg2));
+			"c" (arg2)
+			: "memory");
 
 	return ret;
 }
@@ -46,7 +49,8 @@ static __always_inline long stub_syscall3(long syscall, long arg1, long arg2,
 	long ret;
 
 	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1),
-			"c" (arg2), "d" (arg3));
+			"c" (arg2), "d" (arg3)
+			: "memory");
 
 	return ret;
 }
@@ -57,7 +61,8 @@ static __always_inline long stub_syscall4(long syscall, long arg1, long arg2,
 	long ret;
 
 	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1),
-			"c" (arg2), "d" (arg3), "S" (arg4));
+			"c" (arg2), "d" (arg3), "S" (arg4)
+			: "memory");
 
 	return ret;
 }
@@ -68,7 +73,8 @@ static __always_inline long stub_syscall5(long syscall, long arg1, long arg2,
 	long ret;
 
 	__asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1),
-			"c" (arg2), "d" (arg3), "S" (arg4), "D" (arg5));
+			"c" (arg2), "d" (arg3), "S" (arg4), "D" (arg5)
+			: "memory");
 
 	return ret;
 }
-- 
2.41.0


_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: [PATCH v3 04/11] um: Don't use vfprintf() for os_info()

[Richard Weinberger]

On Fri, Nov 10, 2023 at 12:03 PM <benjamin@sipsolutions.net> wrote:
>
> From: Benjamin Berg <benjamin@sipsolutions.net>
>
> The threads allocated inside the kernel have only a single page of
> stack. Unfortunately, the vfprintf function in standard glibc may use
> too much stack-space, overflowing it.

Another option is giving the helper threads more memory, we don't have
that many.
Did you explore this option already?

-- 
Thanks,
//richard

Re: [PATCH v3 09/11] um: Rely on PTRACE_SETREGSET to set FS/GS base registers

[Richard Weinberger]

On Fri, Nov 10, 2023 at 12:03 PM <benjamin@sipsolutions.net> wrote:
>
> From: Benjamin Berg <benjamin@sipsolutions.net>
>
> These registers are saved/restored together with the other general
> registers using ptrace. In arch_set_tls we then just need to set the
> register and it will be synced back normally.
>
> Most of this logic was introduced in commit f355559cf7845 ("[PATCH] uml:
> x86_64 thread fixes"). However, at least today we can rely on ptrace to

Do you know since when exactly? I don't want to break UML in subtle ways
on old kernels.

-- 
Thanks,
//richard

Re: [PATCH v3 09/11] um: Rely on PTRACE_SETREGSET to set FS/GS base registers

[Richard Weinberger]

On Fri, Jan 5, 2024 at 12:05 AM Richard Weinberger
<richard.weinberger@gmail.com> wrote:
> > Most of this logic was introduced in commit f355559cf7845 ("[PATCH] uml:
> > x86_64 thread fixes"). However, at least today we can rely on ptrace to
>
> Do you know since when exactly? I don't want to break UML in subtle ways
> on old kernels.

BTW: I have applied the series up to here. Thanks a lot for cleaning
up all this. :-)

-- 
Thanks,
//richard

Re: [PATCH v3 04/11] um: Don't use vfprintf() for os_info()

[Benjamin Berg]

Hi,

On Thu, 2024-01-04 at 23:37 +0100, Richard Weinberger wrote:
> On Fri, Nov 10, 2023 at 12:03 PM <benjamin@sipsolutions.net> wrote:
> > 
> > From: Benjamin Berg <benjamin@sipsolutions.net>
> > 
> > The threads allocated inside the kernel have only a single page of
> > stack. Unfortunately, the vfprintf function in standard glibc may use
> > too much stack-space, overflowing it.
> 
> Another option is giving the helper threads more memory, we don't have
> that many.
> Did you explore this option already?

I had not really considered that.

One thing though is that userspace_tramp calls os_info while working
with the stub stack. So that also means setting STUB_DATA_PAGES to 2.
But, I suspect we may want to do that anyway eventually to fit the ever
increasing mcontext size for all the AVX512 registers and such.

As I said, I think that is fine to do.

Benjamin

Re: [PATCH v3 04/11] um: Don't use vfprintf() for os_info()

[Johannes Berg]

On Fri, 2024-01-05 at 09:12 +0100, Benjamin Berg wrote:
> > Another option is giving the helper threads more memory, we don't have
> > that many.
> > Did you explore this option already?
> 
> I had not really considered that.
> 
> One thing though is that userspace_tramp calls os_info while working
> with the stub stack. So that also means setting STUB_DATA_PAGES to 2.
> But, I suspect we may want to do that anyway eventually to fit the ever
> increasing mcontext size for all the AVX512 registers and such.

That'll probably happen eventually regardless ...

> As I said, I think that is fine to do.

But I'm not sure it's a good solution to give more stack space to the
threads and think/hope that glibc won't make more assumptions about the
amount of space it can use ... who knows if it uses alloca() inside
somewhere, for example? After all, userspace stacks are multiple
megabytes by default?

johannes

Re: [PATCH v3 04/11] um: Don't use vfprintf() for os_info()

[Richard Weinberger]

----- Ursprüngliche Mail -----
> Von: "Johannes Berg" <johannes@sipsolutions.net>
> An: "Benjamin Berg" <benjamin@sipsolutions.net>, "Richard Weinberger" <richard.weinberger@gmail.com>
> CC: "linux-um" <linux-um@lists.infradead.org>
> Gesendet: Freitag, 5. Januar 2024 09:56:11
> Betreff: Re: [PATCH v3 04/11] um: Don't use vfprintf() for os_info()

> On Fri, 2024-01-05 at 09:12 +0100, Benjamin Berg wrote:
>> > Another option is giving the helper threads more memory, we don't have
>> > that many.
>> > Did you explore this option already?
>> 
>> I had not really considered that.
>> 
>> One thing though is that userspace_tramp calls os_info while working
>> with the stub stack. So that also means setting STUB_DATA_PAGES to 2.
>> But, I suspect we may want to do that anyway eventually to fit the ever
>> increasing mcontext size for all the AVX512 registers and such.
> 
> That'll probably happen eventually regardless ...
> 
>> As I said, I think that is fine to do.
> 
> But I'm not sure it's a good solution to give more stack space to the
> threads and think/hope that glibc won't make more assumptions about the
> amount of space it can use ... who knows if it uses alloca() inside
> somewhere, for example? After all, userspace stacks are multiple
> megabytes by default?

For thread stacks things are a bit different.

glibc is in general more heap than stack greedy, it uses malloc() almost everywhere.
For pthreads the minimal stack size on x86 is 16k (_SC_THREAD_STACK_MIN).
Donating four pages to each helper thread seems okay to me.
And if I understand _SC_THREAD_STACK_MIN correctly, this is the minimal stack
size the glibc library can deal with.

Thanks,
//richard

Re: [PATCH v3 09/11] um: Rely on PTRACE_SETREGSET to set FS/GS base registers

[Benjamin Berg]

Hi,

On Fri, 2024-01-05 at 00:05 +0100, Richard Weinberger wrote:
> On Fri, Nov 10, 2023 at 12:03 PM <benjamin@sipsolutions.net> wrote:
> > 
> > From: Benjamin Berg <benjamin@sipsolutions.net>
> > 
> > These registers are saved/restored together with the other general
> > registers using ptrace. In arch_set_tls we then just need to set
> > the
> > register and it will be synced back normally.
> > 
> > Most of this logic was introduced in commit f355559cf7845 ("[PATCH]
> > uml:
> > x86_64 thread fixes"). However, at least today we can rely on
> > ptrace to
> 
> Do you know since when exactly? I don't want to break UML in subtle ways
> on old kernels.

To be honest, I don't remember, and I doubt I really understood what I
was doing.

Anyway, I now found this commit now, which is contained in v2.6.25:

commit df5d438e33d7fc914ba9b6e0d6b019a8966c5fcc
Author: Roland McGrath <roland@redhat.com>
Date:   Wed Jan 30 13:30:45 2008 +0100

    x86: ptrace fs/gs_base
    
    The fs_base and gs_base fields are available in user_regs_struct.
    But reading these via ptrace (PTRACE_GETREGS or PTRACE_PEEKUSR) does
    not give a reliably useful value.  The thread_struct fields are 0
    when do_arch_prctl decided to use a GDT slot instead of MSR_FS_BASE,
    which it does for a value under 1<<32.
    
    This changes ptrace access to fs_base and gs_base to work like
    PTRACE_ARCH_PRCTL does.  That is, it reads the base address that
    user-mode memory access using the fs/gs instruction prefixes will
    use, regardless of how it's being implemented in the kernel.  The
    MSR vs GDT is an implementation detail that is pretty much hidden
    from userland in the actual using, and there is no reason that
    ptrace should give the internal implementation picture rather than
    the user-mode semantic picture.  In the case of setting the value,
    this can implicitly change the fsindex/gsindex value (also
    separately in user_regs_struct), which is what happens when the
    thread calls arch_prctl itself.  In a PTRACE_SETREGS, the fs_base
    change will come after the fsindex change due to the order of the
    struct, and so a change the debugger made to fs_base will have the
    effect intended, another part of the user_regs_struct will now
    differ when read back from what the debugger wrote.
    
    This makes PTRACE_ARCH_PRCTL obsolete.  We could consider declaring
    it deprecated and removing it one day, though there is no hurry.
    For the foreseeable future, debuggers have to assume an old kernel
    that does not report reliable fs_base/gs_base values in user_regs_struct
    and stick to PTRACE_ARCH_PRCTL anyway.

I think the last paragraph is quite straight forward in saying that we
do not need anything special anymore. Note that the original commit
adding the code pre-dates this commit by about a year, i.e.:

commit f355559cf78455ed6be103b020e4b800230c64eb
Author: Jeff Dike <jdike@addtoit.com>
Date:   Sat Feb 10 01:44:29 2007 -0800

    [PATCH] uml: x86_64 thread fixes
    
    x86_64 needs some TLS fixes.  What was missing was remembering the child
    thread id during clone and stuffing it into the child during each context
    switch.
    
    The %fs value is stored separately in the thread structure since the host
    controls what effect it has on the actual register file.  The host also needs
    to store it in its own thread struct, so we need the value kept outside the
    register file.
    
    arch_prctl_skas was fixed to call PTRACE_ARCH_PRCTL appropriately.  There is
    some saving and restoring of registers in the ARCH_SET_* cases so that the
    correct set of registers are changed on the host and restored to the process
    when it runs again.

There was some shuffling around in the ptrace code at some point,
eventually switching from doing an arch_prctl call to setting things
more directly. But I don't think that matters to us.

Benjamin

Re: [PATCH v3 09/11] um: Rely on PTRACE_SETREGSET to set FS/GS base registers

[Richard Weinberger]

----- Ursprüngliche Mail -----
> Von: "Benjamin Berg" <benjamin@sipsolutions.net>
> On Fri, 2024-01-05 at 00:05 +0100, Richard Weinberger wrote:
>> On Fri, Nov 10, 2023 at 12:03 PM <benjamin@sipsolutions.net> wrote:
>> > 
>> > From: Benjamin Berg <benjamin@sipsolutions.net>
>> > 
>> > These registers are saved/restored together with the other general
>> > registers using ptrace. In arch_set_tls we then just need to set
>> > the
>> > register and it will be synced back normally.
>> > 
>> > Most of this logic was introduced in commit f355559cf7845 ("[PATCH]
>> > uml:
>> > x86_64 thread fixes"). However, at least today we can rely on
>> > ptrace to
>> 
>> Do you know since when exactly? I don't want to break UML in subtle ways
>> on old kernels.
> 
> To be honest, I don't remember, and I doubt I really understood what I
> was doing.
> 
> Anyway, I now found this commit now, which is contained in v2.6.25:

Okay, 2.6.26 is way older than anything supported. So I'll apply our patch. :-)

Thanks,
//richard