On Tue, 1 Nov 2005, Willy Tarreau wrote:
> But if it's a pointer why don't you declare them unsigned long then ?
> C defines the long as the integer the right size to store a pointer.
^
Is it C?
Since on Wintendo P64 it's not true...
Gr{oetje,eeting}s,
On Mon, Oct 31, 2005 at 01:14:22PM -0800, Matt Mackall wrote:
> On Mon, Oct 31, 2005 at 10:05:15PM +0100, Geert Uytterhoeven wrote:
> > On Mon, 31 Oct 2005, Matt Mackall wrote:
> > > inflate: remove legacy type definitions from callers
> > >
> > > This replaces the legacy zlib typedefs and usage w
On Tue, Nov 01, 2005 at 06:09:00AM +0100, Christoph Hellwig wrote:
> [Let's try again now that sys_ptrace returns long everywhere mainline..]
>
> The sys_ptrace boilerplate code (everything outside the big switch
> statement for the arch-specific requests) is shared by most
> architectures. This
[Let's try again now that sys_ptrace returns long everywhere mainline..]
The sys_ptrace boilerplate code (everything outside the big switch
statement for the arch-specific requests) is shared by most
architectures. This patch moves it to kernel/ptrace.c and leaves the
arch-specific code as arch_p
On Tue, Nov 01, 2005 at 11:24:27AM +1100, Paul Mackerras wrote:
> Matt,
>
> My concern about this series of patches is that it will make it harder
> to keep the kernel zlib in sync with the upstream zlib.
This code is very long out of sync with the existing upstream zlib in
terms of coding style
Matt,
My concern about this series of patches is that it will make it harder
to keep the kernel zlib in sync with the upstream zlib.
Are you signing up to track the upstream zlib and apply any changes
made there to the kernel version, for the forseeable future?
Paul.
On Mon, Oct 31, 2005 at 03:11:21PM -0800, Matt Mackall wrote:
> On Mon, Oct 31, 2005 at 11:10:52PM +, Russell King wrote:
> > That's what threading is for. 8)
>
> What's what's threading is for?
>
> > > I think for ARM, we can simply do -DINITDATA=const, yes?
> >
> > No, unless you want to m
On Mon, Oct 31, 2005 at 11:10:52PM +, Russell King wrote:
> That's what threading is for. 8)
What's what's threading is for?
> > I think for ARM, we can simply do -DINITDATA=const, yes?
>
> No, unless you want to make this const:
>
> -static u8 window[0x8000]; /* use a statically allocated
On Mon, Oct 31, 2005 at 03:02:33PM -0800, Matt Mackall wrote:
> On Mon, Oct 31, 2005 at 10:45:01PM +, Russell King wrote:
> > On Mon, Oct 31, 2005 at 02:54:52PM -0600, Matt Mackall wrote:
> > > inflate: remove include of lib/inflate.c and use proper linking
> > >
> > > - make free_mem_ptr vars
On Mon, Oct 31, 2005 at 02:57:46PM -0800, Matt Mackall wrote:
> On Mon, Oct 31, 2005 at 10:43:01PM +, Russell King wrote:
> > On Mon, Oct 31, 2005 at 02:54:51PM -0600, Matt Mackall wrote:
> > > inflate: mark some arrays as INITDATA and define it in in-core callers
> >
> > This breaks ARM. Our
On Mon, Oct 31, 2005 at 10:45:01PM +, Russell King wrote:
> On Mon, Oct 31, 2005 at 02:54:52PM -0600, Matt Mackall wrote:
> > inflate: remove include of lib/inflate.c and use proper linking
> >
> > - make free_mem_ptr vars nonstatic
> > - make gunzip nonstatic
> > - add gunzip prototype to new
On Mon, Oct 31, 2005 at 10:43:01PM +, Russell King wrote:
> On Mon, Oct 31, 2005 at 02:54:51PM -0600, Matt Mackall wrote:
> > inflate: mark some arrays as INITDATA and define it in in-core callers
>
> This breaks ARM. Our decompressor has some rather odd requirements
> due to the way we suppo
On Mon, Oct 31, 2005 at 02:54:52PM -0600, Matt Mackall wrote:
> inflate: remove include of lib/inflate.c and use proper linking
>
> - make free_mem_ptr vars nonstatic
> - make gunzip nonstatic
> - add gunzip prototype to new inflate.h
> - add per-arch Makefile bits
> - change inflate.c includes to
On Mon, Oct 31, 2005 at 02:54:51PM -0600, Matt Mackall wrote:
> inflate: mark some arrays as INITDATA and define it in in-core callers
This breaks ARM. Our decompressor has some rather odd requirements
due to the way we support PIC - it's PIC text with fixed data.
This means that all fixed initi
On Mon, Oct 31, 2005 at 10:05:15PM +0100, Geert Uytterhoeven wrote:
> On Mon, 31 Oct 2005, Matt Mackall wrote:
> > inflate: remove legacy type definitions from callers
> >
> > This replaces the legacy zlib typedefs and usage with kernel types in
> > all the inflate users.
>
> > -static ulg free_m
inflate: make in-core inflate share common CRC code
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/lib/inflate.c
===
--- 2.6.14.orig/lib/inflate.c 2005-10-28 22:04:32.0 -0700
+++ 2.6.14/lib/inflate.c2
inflate: replace call to memzero with simple loop
This is the only user of memzero in the inflate code and it's only for
16 bytes. Removing this lets us drop a copy of memzero from most
lib/inflate users.
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/lib/inflate.c
===
inflate: remove include of lib/inflate.c and use proper linking
- make free_mem_ptr vars nonstatic
- make gunzip nonstatic
- add gunzip prototype to new inflate.h
- add per-arch Makefile bits
- change inflate.c includes to inflate.h includes
- change NO_INFLATE_MALLOC to CORE
- compile core kernel
inflate: mark some arrays as INITDATA and define it in in-core callers
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: tiny/init/do_mounts_rd.c
===
--- tiny.orig/init/do_mounts_rd.c 2005-09-30 23:45:21.0 -0700
++
inflate: remove legacy DEBG macros
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/lib/inflate.c
===
--- 2.6.14.orig/lib/inflate.c 2005-10-28 22:04:23.0 -0700
+++ 2.6.14/lib/inflate.c2005-10-28 22:04:2
inflate: replace get_byte with readbyte
Each inflate user was providing a get_byte macro that hid the details
of tracking the input buffer. This is now handled with new variables
in the iostate structure and a callback (most users pass NULL as the
entire input is provided in a single buffer).
Sig
inflate: lindent and general reformatting
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: tiny/lib/inflate.c
===
--- tiny.orig/lib/inflate.c 2005-09-27 19:31:03.0 -0700
+++ tiny/lib/inflate.c 2005-09-27 19:58:58.0
inflate: refactor inflate malloc code
Inflate requires some dynamic memory allocation very early in the boot
process and this is provided with a set of four functions:
malloc/free/gzip_mark/gzip_release.
The old inflate code used a mark/release strategy rather than
implement free. This new versio
inflate: cleanup input logic
Transform ugly macros to inlines
Kill mask_bits table
Eliminate magic underrun handling (dealt with by getbyte())
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: tiny/lib/inflate.c
===
--- tiny.or
On Mon, 31 Oct 2005, Matt Mackall wrote:
> inflate: remove legacy type definitions from callers
>
> This replaces the legacy zlib typedefs and usage with kernel types in
> all the inflate users.
> -static ulg free_mem_ptr;
> -static ulg free_mem_ptr_end;
> +static u32 free_mem_ptr;
> +static u32
inflate: error handling cleanup
This passes the error function as a callback rather than using a
global symbol.
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/arch/alpha/boot/misc.c
===
--- 2.6.14.orig/arch/alpha/boot
inflate: internalize most of the output window handling
This moves the inflate window definition and handling into
lib/inflate.c and rather than calling a global flush_window function,
now calls a flush callback. The callback in most callers is also
greatly simplified.
Signed-off-by: Matt Mackall
inflate: remove a bunch of declarations/definitions from callers
This removes:
- memset/memzero/memcpy implementations
- OF
- STATIC
- gzip flag byte defines
- unused debug defines
and saves an average of 50 lines in each of 12 users.
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.
inflate: move CRC calculation
Each inflate user was doing its own open-coded CRC calculation and
initializing its own CRC table. This is now hidden inside
lib/inflate.c
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/arch/alpha/boot/misc.c
==
inflate: small constant tidy-up
Index: 2.6.14/lib/inflate.c
===
--- 2.6.14.orig/lib/inflate.c 2005-10-28 22:04:29.0 -0700
+++ 2.6.14/lib/inflate.c2005-10-28 22:04:31.0 -0700
@@ -52,7 +52,7 @@
The Huffman
inflate: cleanup CRC calculation
Move CRC calculation into inflate code
Cleanup table calculation code
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/lib/inflate.c
===
--- 2.6.14.orig/lib/inflate.c 2005-10-28 22:01:
inflate: kill legacy bits
Kill RCSID
Kill old includes and defines
Kill screwy types
Kill unused memory usage tracking
Kill 'register' usage
Kill unused tracing calls
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: tiny/lib/inflate.c
===
inflate: move globals into a state structure
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/lib/inflate.c
===
--- 2.6.14.orig/lib/inflate.c 2005-10-28 20:41:17.0 -0700
+++ 2.6.14/lib/inflate.c2005-10-
inflate: tidy user declarations
Regroup the few remaining gzip-related declarations in users. As the
bulk of the users are now _not_ a collection of routines copied from
gzip, that comment is removed.
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/arch/alpha/boot/misc.c
==
inflate: cleanup huffman table code
sensible names for huft struct members
get rid of assignment-in-expression usage
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/lib/inflate.c
===
--- 2.6.14.orig/lib/inflate.c 200
inflate: remove legacy type definitions from callers
This replaces the legacy zlib typedefs and usage with kernel types in
all the inflate users.
Signed-off-by: Matt Mackall <[EMAIL PROTECTED]>
Index: 2.6.14/arch/alpha/boot/misc.c
=
36 matches
Mail list logo