Skip to contentSkip to conversation

x86: Drop 32-bit support ... finally.

PATCH In review Borislav Petkov 8 participants 11 messages Apr 1, 2015

Subsystems (2)
lore
Share & tools
lore

Borislav Petkov

author maintainer
#0 lore raw

PATCH 1/1 x86: Drop 32-bit support ... finally.

From: Borislav Petkov <[email protected]>

Today, April 1st 2015, marks almost 15 years since the introduction of the 64-bit extensions to the x86 architecture. And frankly, 15 years was a graceful period enough for people to move to 64-bit. Therefore, today, I'm removing 32-bit support from x86 Linux. And it was about friggin' time...

32-bit code is old, full of nasty hacks and keeps always breaking when we do our shiny new features for 64-bit. And frankly, no one cares about 32-bit. If you do, then you're wrong and you need to get with the program. Go out, take a deep breath, go for a walk and the first thing you do when you come back is *take* *a* *look* at the goddam calendar. Time to say goodbye to your old 32-bit sand. Get a 64-bit processor. Live a little. Come to the real world.

This is a minimal patch which prevents from building a 32-bit kernel only. Removing of the rest of the crazy ifdeffery and potential cleanups needs to come later, in nice clean patches. This is especially true for you, you patch count wankers. Restrain yourself!

Cc H. Peter Anvin <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Tony Luck <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Matt Fleming <[email protected]>
Cc Jörg Rödel <[email protected]>
Cc Jiri Kosina <[email protected]>
Cc Jeff Mahoney <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Vlastimil Babka <[email protected]>
Cc Andre Przywara <[email protected]>
Cc Andreas Herrmann <[email protected]>
Cc Robert Richter <[email protected]>
Cc Richard Weinberger <[email protected]>
---
 Makefile         | 3 ---
 arch/x86/Kconfig | 9 ++-------
 2 files changed, 2 insertions(+), 10 deletions(-)
Makefile +0 -3
index da36a3be7969..0a907dd934b6 100644
@@ -256,9 +256,6 @@ UTS_MACHINE 	:= $(ARCH)
 SRCARCH 	:= $(ARCH)
 
 # Additional ARCH settings for x86
-ifeq ($(ARCH),i386)
-        SRCARCH := x86
-endif
 ifeq ($(ARCH),x86_64)
         SRCARCH := x86
 endif
arch/x86/Kconfig +2 -7
index 8db84b0e5bbd..9fed79e3ed39 100644
@@ -1,13 +1,8 @@
-# Select 32 or 64 bit
 config 64BIT
-	bool "64-bit kernel" if ARCH = "x86"
-	default ARCH != "i386"
-	---help---
-	  Say yes to build a 64-bit kernel - formerly known as x86_64
-	  Say no to build a 32-bit kernel - formerly known as i386
+	def_bool y
 
 config X86_32
-	def_bool y
+	def_bool n
 	depends on !64BIT
 	select CLKSRC_I8253
 	select HAVE_UID16
Signature · 1 line
2.3.3
author maintainer
#1 lore raw

peterz says I forgot to Cc Linus. My bad. Fixed.

Quoted 82 lines · Borislav Petkov

From: Borislav Petkov <[email protected]>

Today, April 1st 2015, marks almost 15 years since the introduction of the 64-bit extensions to the x86 architecture. And frankly, 15 years was a graceful period enough for people to move to 64-bit. Therefore, today, I'm removing 32-bit support from x86 Linux. And it was about friggin' time...

32-bit code is old, full of nasty hacks and keeps always breaking when we do our shiny new features for 64-bit. And frankly, no one cares about 32-bit. If you do, then you're wrong and you need to get with the program. Go out, take a deep breath, go for a walk and the first thing you do when you come back is *take* *a* *look* at the goddam calendar. Time to say goodbye to your old 32-bit sand. Get a 64-bit processor. Live a little. Come to the real world.

This is a minimal patch which prevents from building a 32-bit kernel only. Removing of the rest of the crazy ifdeffery and potential cleanups needs to come later, in nice clean patches. This is especially true for you, you patch count wankers. Restrain yourself!

Cc H. Peter Anvin <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Tony Luck <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Matt Fleming <[email protected]>
Cc Jörg Rödel <[email protected]>
Cc Jiri Kosina <[email protected]>
Cc Jeff Mahoney <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Vlastimil Babka <[email protected]>
Cc Andre Przywara <[email protected]>
Cc Andreas Herrmann <[email protected]>
Cc Robert Richter <[email protected]>
Cc Richard Weinberger <[email protected]>
---
 Makefile         | 3 ---
 arch/x86/Kconfig | 9 ++-------
 2 files changed, 2 insertions(+), 10 deletions(-)
Makefile +0 -3
index da36a3be7969..0a907dd934b6 100644
@@ -256,9 +256,6 @@ UTS_MACHINE 	:= $(ARCH)
 SRCARCH 	:= $(ARCH)
 
 # Additional ARCH settings for x86
-ifeq ($(ARCH),i386)
-        SRCARCH := x86
-endif
 ifeq ($(ARCH),x86_64)
         SRCARCH := x86
 endif
arch/x86/Kconfig +2 -7
index 8db84b0e5bbd..9fed79e3ed39 100644
@@ -1,13 +1,8 @@
-# Select 32 or 64 bit
 config 64BIT
-	bool "64-bit kernel" if ARCH = "x86"
-	default ARCH != "i386"
-	---help---
-	  Say yes to build a 64-bit kernel - formerly known as x86_64
-	  Say no to build a 32-bit kernel - formerly known as i386
+	def_bool y
 
 config X86_32
-	def_bool y
+	def_bool n
 	depends on !64BIT
 	select CLKSRC_I8253
 	select HAVE_UID16
Signature · 1 line
2.3.3
Signature · 5 lines
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--
#2 lore raw
Quoted 1 line · Borislav Petkov

peterz says I forgot to Cc Linus. My bad. Fixed.

Yeah, although this cleanup really shouldn't be controversial so I'll merge it up for a v4.0 merge ASAP.

Thanks,

Ingo

#3 lore raw
In reply to Borislav Petkov

(cc'ing Andi)

Quoted 8 lines · Borislav Petkov
Quoted 7 lines

From: Borislav Petkov <[email protected]>

Today, April 1st 2015, marks almost 15 years since the introduction of the 64-bit extensions to the x86 architecture. And frankly, 15 years was a graceful period enough for people to move to 64-bit. Therefore, today, I'm removing 32-bit support from x86 Linux. And it was about friggin' time...

No, I vaguely remember booting the first 64 bit cpu (on real hw) for the 2002-2003 time frame. I need to check my memories on a more exact date, but you guys from Suse should know better. So 15 years are not yet over and I really would like to wait at least until then. Another question is if 64-bit simulators take into account here.

Though, I assume a discussion on this topic will at least take 1-2 years, so it's good to start with it now. Also, I really would like to finally switch off and retire my K7 which still runs some regression tests (will need to check the actual used kernel version). It was a good time, but now I realize it's somehow over. Considering this and apart from my concerns above:

-Robert

#4 lore raw
Quoted 11 lines · Robert Richter

(cc'ing Andi)

Quoted 8 lines · Borislav Petkov
Quoted 7 lines

From: Borislav Petkov <[email protected]>

Today, April 1st 2015, marks almost 15 years since the introduction of the 64-bit extensions to the x86 architecture. And frankly, 15 years was a graceful period enough for people to move to 64-bit. Therefore, today, I'm removing 32-bit support from x86 Linux. And it was about friggin' time...

Quoted 3 lines

No, I vaguely remember booting the first 64 bit cpu (on real hw) for the 2002-2003 time frame. I need to check my memories on a more exact date, but you guys from Suse should know better. So 15 years are not

Looks like Feb 7 2002, assuming this is genuine:

http://www.theinquirer.net/inquirer/news/1049122/amd-microsoft-hammer-memo-leaked

Not sure about Linux first booting 64 bit.

-Robert

Quoted 13 lines

yet over and I really would like to wait at least until then. Another question is if 64-bit simulators take into account here.

Though, I assume a discussion on this topic will at least take 1-2 years, so it's good to start with it now. Also, I really would like to finally switch off and retire my K7 which still runs some regression tests (will need to check the actual used kernel version). It was a good time, but now I realize it's somehow over. Considering this and apart from my concerns above:

-Robert

#6 lore raw
In reply to Robert Richter
Quoted 10 lines · Robert Richter

[...]

Though, I assume a discussion on this topic will at least take 1-2 years, so it's good to start with it now. Also, I really would like to finally switch off and retire my K7 which still runs some regression tests (will need to check the actual used kernel version). It was a good time, but now I realize it's somehow over. Considering this and apart from my concerns above:

Yeah, and phasing out 32-bit support will make it easier to add the (inevitable) 128-bit support as well, lifting the ~18 exabytes memory limit. That will allow us to move the 64-bit code into 'bugfix only' legacy status.

We really don't want to have this unmaintainable mixture of:

128-bit support
64-bit support (legacy mode)
32-bit support (compat mode)
16-bit support (vm86 mode)
8-bit support (UART mode)

... all in the same kernel! We want a cleaner, more maintainable structure:

	128-bit support
	 64-bit support (compat mode)

... which more closely resembles today's kernel.

Another related change I'd like to make is to drop !SMP support and to default the kernel build to 4096 CPUs, and to move lower CPU counts to under CONFIG_EXPERT=y. By the time the commit trickles through to distros 1000-4000 core phones will be commonplace, so we might as well start the transition today.

Thanks,

Ingo

#7 lore raw
Quoted 9 lines

We really don't want to have this unmaintainable mixture of:

128-bit support
64-bit support (legacy mode)
32-bit support (compat mode)
16-bit support (vm86 mode)
8-bit support (UART mode)

... all in the same kernel!

I think a lot of us working on other free Unixlike OS platforms have reached a similar conclusion. It's a natural segmentation. The final commit should include the following helpful table

	Bits		Free Unixlike OS
        ---------------------------------
	8bit		FUZIX
	16bit		ELKS
	32bit		NetBSD/FreeBSD
	64bit		Linux

(For 18 and 36bit consult a specialist.. probably a medical one)

Alan

#8 lore raw
Quoted 6 lines · One Thousand Gnomes

I think a lot of us working on other free Unixlike OS platforms have reached a similar conclusion. It's a natural segmentation. The final commit should include the following helpful table

Quoted 8 lines
	Bits		Free Unixlike OS
        ---------------------------------
	8bit		FUZIX
	16bit		ELKS
	32bit		NetBSD/FreeBSD
	64bit		Linux

(For 18 and 36bit consult a specialist.. probably a medical one)

Hey! My father still has a PDP-15/30, complete with core memory, 4 DECtape units, and a high speed punched paper tape reader, all in a really tiny footprint of 4 full-height 19 inch racks, in his house. It hasn't been turned on in over a decade, but doesn't deserve a Free Unixlike OS of its own? :-)

- Ted

#9 lore raw
Quoted 5 lines

Hey! My father still has a PDP-15/30, complete with core memory, 4 DECtape units, and a high speed punched paper tape reader, all in a really tiny footprint of 4 full-height 19 inch racks, in his house. It hasn't been turned on in over a decade, but doesn't deserve a Free Unixlike OS of its own? :-)

It would be greener to write a java emulator to run in a browser on a phone (1e4 less power? 1e5??) Probably run 100x faster too.

-Tony

#10 lore raw
Quoted 8 lines · Tony
Quoted 5 lines

Hey! My father still has a PDP-15/30, complete with core memory, 4 DECtape units, and a high speed punched paper tape reader, all in a really tiny footprint of 4 full-height 19 inch racks, in his house. It hasn't been turned on in over a decade, but doesn't deserve a Free Unixlike OS of its own? :-)

It would be greener to write a java emulator to run in a browser on a phone (1e4 less power? 1e5??) Probably run 100x faster too.

I wonder at what point do we say the same thing about ia64 ?

Dave