These two patches remove the last two in-tree users of exec domains.
Both seem to be abandoned and historical.
Russel, Tony, are you fine with that?
The plan is to get rid of exec domains at all. IOW no more register_exec_domain,
no signal translation.
I have a follow up series for that already in my local tree.
Thanks,
//richard
[PATCH 1/2] arm: Remove RISC OS personality
[PATCH 2/2] ia64: Remove Linux/x86 domain support
arch/arm/kernel/arthur.c | 94 ------------------------------------
arch/arm/Kconfig | 12 ----
arch/arm/configs/badge4_defconfig | 1
arch/arm/kernel/Makefile | 1
arch/ia64/mm/init.c | 25 ---------
5 files changed, 133 deletions(-)
The RISC OS personality seems to be unused and untested for a long time.
Let's rip it out.
Signed-off-by: Richard Weinberger <[email protected]>
---
arch/arm/Kconfig | 12 -----
arch/arm/configs/badge4_defconfig | 1 -
arch/arm/kernel/Makefile | 1 -
arch/arm/kernel/arthur.c | 94 ---------------------------------------
4 files changed, 108 deletions(-)
delete mode 100644 arch/arm/kernel/arthur.c
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index 245058b..b57afd3 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -2177,18 +2177,6 @@ menu "Userspace binary formats"
source "fs/Kconfig.binfmt"
-config ARTHUR
- tristate "RISC OS personality"
- depends on !AEABI
- help
- Say Y here to include the kernel code necessary if you want to run
- Acorn RISC OS/Arthur binaries under Linux. This code is still very
- experimental; if this sounds frightening, say N and sleep in peace.
- You can also say M here to compile this support as a module (which
- will be called arthur).
-
-endmenu
-
menu "Power management options"
source "kernel/power/Kconfig"
diff --git a/arch/arm/configs/badge4_defconfig b/arch/arm/configs/badge4_defconfig
index 0494c8f..d590098 100644
--- a/arch/arm/configs/badge4_defconfig
+++ b/arch/arm/configs/badge4_defconfig
@@ -12,7 +12,6 @@ CONFIG_CPU_FREQ_GOV_PERFORMANCE=y
CONFIG_FPE_NWFPE=y
CONFIG_BINFMT_AOUT=m
CONFIG_BINFMT_MISC=m
-CONFIG_ARTHUR=m
CONFIG_NET=y
CONFIG_PACKET=y
CONFIG_UNIX=y
diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f..742e076 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -35,7 +35,6 @@ obj-$(CONFIG_CPU_IDLE) += cpuidle.o
obj-$(CONFIG_ISA_DMA_API) += dma.o
obj-$(CONFIG_FIQ) += fiq.o fiqasm.o
obj-$(CONFIG_MODULES) += armksyms.o module.o
-obj-$(CONFIG_ARTHUR) += arthur.o
obj-$(CONFIG_ISA_DMA) += dma-isa.o
obj-$(CONFIG_PCI) += bios32.o isa.o
obj-$(CONFIG_ARM_CPU_SUSPEND) += sleep.o suspend.o
diff --git a/arch/arm/kernel/arthur.c b/arch/arm/kernel/arthur.c
deleted file mode 100644
index 321c529..0000000
--- a/arch/arm/kernel/arthur.c
+++ /dev/null
@@ -1,94 +0,0 @@
-/*
- * linux/arch/arm/kernel/arthur.c
- *
- * Copyright (C) 1998, 1999, 2000, 2001 Philip Blundell
- *
- * Arthur personality
- */
-
-/*
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License
- * as published by the Free Software Foundation; either version
- * 2 of the License, or (at your option) any later version.
- */
-
-#include <linux/module.h>
-#include <linux/personality.h>
-#include <linux/stddef.h>
-#include <linux/signal.h>
-#include <linux/init.h>
-#include <linux/sched.h>
-
-#include <asm/ptrace.h>
-
-/* Arthur doesn't have many signals, and a lot of those that it does
- have don't map easily to any Linux equivalent. Never mind. */
-
-#define ARTHUR_SIGABRT 1
-#define ARTHUR_SIGFPE 2
-#define ARTHUR_SIGILL 3
-#define ARTHUR_SIGINT 4
-#define ARTHUR_SIGSEGV 5
-#define ARTHUR_SIGTERM 6
-#define ARTHUR_SIGSTAK 7
-#define ARTHUR_SIGUSR1 8
-#define ARTHUR_SIGUSR2 9
-#define ARTHUR_SIGOSERROR 10
-
-static unsigned long arthur_to_linux_signals[32] = {
- 0, 1, 2, 3, 4, 5, 6, 7,
- 8, 9, 10, 11, 12, 13, 14, 15,
- 16, 17, 18, 19, 20, 21, 22, 23,
- 24, 25, 26, 27, 28, 29, 30, 31
-};
-
-static unsigned long linux_to_arthur_signals[32] = {
- 0, -1, ARTHUR_SIGINT, -1,
- ARTHUR_SIGILL, 5, ARTHUR_SIGABRT, 7,
- ARTHUR_SIGFPE, 9, ARTHUR_SIGUSR1, ARTHUR_SIGSEGV,
- ARTHUR_SIGUSR2, 13, 14, ARTHUR_SIGTERM,
- 16, 17, 18, 19,
- 20, 21, 22, 23,
- 24, 25, 26, 27,
- 28, 29, 30, 31
-};
-
-static void arthur_lcall7(int nr, struct pt_regs *regs)
-{
- struct siginfo info;
- info.si_signo = SIGSWI;
- info.si_errno = nr;
- /* Bounce it to the emulator */
- send_sig_info(SIGSWI, &info, current);
-}
-
-static struct exec_domain arthur_exec_domain = {
- .name = "Arthur",
- .handler = arthur_lcall7,
- .pers_low = PER_RISCOS,
- .pers_high = PER_RISCOS,
- .signal_map = arthur_to_linux_signals,
- .signal_invmap = linux_to_arthur_signals,
- .module = THIS_MODULE,
-};
-
-/*
- * We could do with some locking to stop Arthur being removed while
- * processes are using it.
- */
-
-static int __init arthur_init(void)
-{
- return register_exec_domain(&arthur_exec_domain);
-}
-
-static void __exit arthur_exit(void)
-{
- unregister_exec_domain(&arthur_exec_domain);
-}
-
-module_init(arthur_init);
-module_exit(arthur_exit);
-
-MODULE_LICENSE("GPL");
--
2.0.1
ia64 exec domain support seems to be incomplete and unused.
It has no custom handler nor a custom signal map.
All it does is showing up in /proc/execdomains.
Let's rip it out.
Signed-off-by: Richard Weinberger <[email protected]>
---
arch/ia64/mm/init.c | 25 -------------------------
1 file changed, 25 deletions(-)
diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index 25c3502..d4a6b79 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -660,31 +660,6 @@ int arch_remove_memory(u64 start, u64 size)
#endif
#endif
-/*
- * Even when CONFIG_IA32_SUPPORT is not enabled it is
- * useful to have the Linux/x86 domain registered to
- * avoid an attempted module load when emulators call
- * personality(PER_LINUX32). This saves several milliseconds
- * on each such call.
- */
-static struct exec_domain ia32_exec_domain;
-
-static int __init
-per_linux32_init(void)
-{
- ia32_exec_domain.name = "Linux/x86";
- ia32_exec_domain.handler = NULL;
- ia32_exec_domain.pers_low = PER_LINUX32;
- ia32_exec_domain.pers_high = PER_LINUX32;
- ia32_exec_domain.signal_map = default_exec_domain.signal_map;
- ia32_exec_domain.signal_invmap = default_exec_domain.signal_invmap;
- register_exec_domain(&ia32_exec_domain);
-
- return 0;
-}
-
-__initcall(per_linux32_init);
-
/**
* show_mem - give short summary of memory stats
*
--
2.0.1
On Sun, Jul 13, 2014 at 7:58 AM, Richard Weinberger <[email protected]> wrote:
Did you read the comment?
> -/*
> - * Even when CONFIG_IA32_SUPPORT is not enabled it is
> - * useful to have the Linux/x86 domain registered to
> - * avoid an attempted module load when emulators call
> - * personality(PER_LINUX32). This saves several milliseconds
> - * on each such call.
> - */
I don't know if there are still people using ia32el to run x86 binaries
on ia64 ... but if there are, they may be sad at the extra delays.
-Tony
Am 17.07.2014 01:20, schrieb Tony Luck:
> On Sun, Jul 13, 2014 at 7:58 AM, Richard Weinberger <[email protected]> wrote:
>
> Did you read the comment?
Sure, please see my comment below.
>> -/*
>> - * Even when CONFIG_IA32_SUPPORT is not enabled it is
>> - * useful to have the Linux/x86 domain registered to
>> - * avoid an attempted module load when emulators call
>> - * personality(PER_LINUX32). This saves several milliseconds
>> - * on each such call.
>> - */
>
> I don't know if there are still people using ia32el to run x86 binaries
> on ia64 ... but if there are, they may be sad at the extra delays.
As written in 0/, the plan is to get rid of exec domains at all.
Therefore personality(PER_LINUX32) would no longer try to request a module.
Thanks,
//richard
On Sun, Jul 13, 2014 at 04:58:57PM +0200, Richard Weinberger wrote:
> The RISC OS personality seems to be unused and untested for a long time.
> Let's rip it out.
>
> Signed-off-by: Richard Weinberger <[email protected]>
Thanks.
Acked-by: Russell King <[email protected]>
--
FTTC broadband for 0.8mile line: currently at 9.5Mbps down 400kbps up
according to speedtest.net.