* [PATCH -v3] flat: fix data sections alignment @ 2009-03-05 17:00 Oskar Schirmer 2009-03-10 7:33 ` Paul Mundt 0 siblings, 1 reply; 14+ messages in thread From: Oskar Schirmer @ 2009-03-05 17:00 UTC (permalink / raw) To: Andrew Morton Cc: Johannes Weiner, David Howells, Russell King, Bryan Wu, Geert Uytterhoeven, Mike Frysinger, Greg Ungerer, Paul Mundt, linux-kernel The flat loader uses an architecture's flat_stack_align() to align the stack but assumes word-alignment is enough for the data sections. However, on the Xtensa S6000 we have registers up to 128bit width which can be used from userspace and therefor need userspace stack and data-section alignment of at least this size. This patch drops flat_stack_align() and uses the same alignment that is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's not defined by the architecture. It also fixes m32r which was obviously kaput, aligning an uninitialized stack entry instead of the stack pointer. Signed-off-by: Oskar Schirmer <os@emlix.com> Signed-off-by: Johannes Weiner <jw@emlix.com> Cc: David Howells <dhowells@redhat.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Bryan Wu <cooloney@kernel.org> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Paul Mundt <lethal@linux-sh.org> Cc: Greg Ungerer <gerg@uclinux.org> Cc: Mike Frysinger <vapier.adi@gmail.com> --- arch/arm/include/asm/flat.h | 3 -- arch/blackfin/include/asm/flat.h | 1 - arch/h8300/include/asm/flat.h | 1 - arch/m68k/include/asm/flat.h | 1 - arch/sh/include/asm/flat.h | 1 - fs/binfmt_flat.c | 46 +++++++++++++++++++++++++------------ include/asm-m32r/flat.h | 1 - 7 files changed, 31 insertions(+), 23 deletions(-) v3: - checkpatch-clean (Andrew Morton) - fixed the alignment space upon allocation calculation (Mike Frysinger) - added a comment for the use of ARCH_SLAB_MINALIGN (Paul Mundt) v2: - replace local flat_data_align by ARCH_SLAB_MINALIGN (Paul Mundt) diff --git a/arch/arm/include/asm/flat.h b/arch/arm/include/asm/flat.h index 1d77e51..59426a4 100644 --- a/arch/arm/include/asm/flat.h +++ b/arch/arm/include/asm/flat.h @@ -5,9 +5,6 @@ #ifndef __ARM_FLAT_H__ #define __ARM_FLAT_H__ -/* An odd number of words will be pushed after this alignment, so - deliberately misalign the value. */ -#define flat_stack_align(sp) sp = (void *)(((unsigned long)(sp) - 4) | 4) #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff --git a/arch/blackfin/include/asm/flat.h b/arch/blackfin/include/asm/flat.h index e70074e..733a178 100644 --- a/arch/blackfin/include/asm/flat.h +++ b/arch/blackfin/include/asm/flat.h @@ -10,7 +10,6 @@ #include <asm/unaligned.h> -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) diff --git a/arch/h8300/include/asm/flat.h b/arch/h8300/include/asm/flat.h index 2a87350..bd12b31 100644 --- a/arch/h8300/include/asm/flat.h +++ b/arch/h8300/include/asm/flat.h @@ -5,7 +5,6 @@ #ifndef __H8300_FLAT_H__ #define __H8300_FLAT_H__ -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) 1 #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff --git a/arch/m68k/include/asm/flat.h b/arch/m68k/include/asm/flat.h index 814b517..a0e2907 100644 --- a/arch/m68k/include/asm/flat.h +++ b/arch/m68k/include/asm/flat.h @@ -5,7 +5,6 @@ #ifndef __M68KNOMMU_FLAT_H__ #define __M68KNOMMU_FLAT_H__ -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff --git a/arch/sh/include/asm/flat.h b/arch/sh/include/asm/flat.h index d3b2b4f..5d84df5 100644 --- a/arch/sh/include/asm/flat.h +++ b/arch/sh/include/asm/flat.h @@ -12,7 +12,6 @@ #ifndef __ASM_SH_FLAT_H #define __ASM_SH_FLAT_H -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff --git a/fs/binfmt_flat.c b/fs/binfmt_flat.c index 5cebf0b..697f6b5 100644 --- a/fs/binfmt_flat.c +++ b/fs/binfmt_flat.c @@ -41,6 +41,7 @@ #include <asm/uaccess.h> #include <asm/unaligned.h> #include <asm/cacheflush.h> +#include <asm/page.h> /****************************************************************************/ @@ -54,6 +55,18 @@ #define DBG_FLT(a...) #endif +/* + * User data (stack, data section and bss) needs to be aligned + * for the same reasons as SLAB memory is, and to the same amount. + * Avoid duplicating architecture specific code by using the same + * macro as with SLAB allocation: + */ +#ifdef ARCH_SLAB_MINALIGN +#define FLAT_DATA_ALIGN (ARCH_SLAB_MINALIGN) +#else +#define FLAT_DATA_ALIGN (sizeof(void *)) +#endif + #define RELOC_FAILED 0xff00ff01 /* Relocation incorrect somewhere */ #define UNLOADED_LIB 0x7ff000ff /* Placeholder for unused library */ @@ -114,20 +127,18 @@ static unsigned long create_flat_tables( int envc = bprm->envc; char uninitialized_var(dummy); - sp = (unsigned long *) ((-(unsigned long)sizeof(char *))&(unsigned long) p); + sp = (unsigned long *)p; + sp -= (envc + argc + 2) + 1 + (flat_argvp_envp_on_stack() ? 2 : 0); + sp = (unsigned long *) ((unsigned long)sp & -FLAT_DATA_ALIGN); + argv = sp + 1 + (flat_argvp_envp_on_stack() ? 2 : 0); + envp = argv + (argc + 1); - sp -= envc+1; - envp = sp; - sp -= argc+1; - argv = sp; - - flat_stack_align(sp); if (flat_argvp_envp_on_stack()) { - --sp; put_user((unsigned long) envp, sp); - --sp; put_user((unsigned long) argv, sp); + put_user((unsigned long) envp, sp + 2); + put_user((unsigned long) argv, sp + 1); } - put_user(argc,--sp); + put_user(argc, sp); current->mm->arg_start = (unsigned long) p; while (argc-->0) { put_user((unsigned long) p, argv++); @@ -558,7 +569,9 @@ static int load_flat_file(struct linux_binprm * bprm, ret = realdatastart; goto err; } - datapos = realdatastart + MAX_SHARED_LIBS * sizeof(unsigned long); + datapos = ALIGN(realdatastart + + MAX_SHARED_LIBS * sizeof(unsigned long), + FLAT_DATA_ALIGN); DBG_FLT("BINFMT_FLAT: Allocated data+bss+stack (%d bytes): %x\n", (int)(data_len + bss_len + stack_len), (int)datapos); @@ -604,9 +617,12 @@ static int load_flat_file(struct linux_binprm * bprm, } realdatastart = textpos + ntohl(hdr->data_start); - datapos = realdatastart + MAX_SHARED_LIBS * sizeof(unsigned long); - reloc = (unsigned long *) (textpos + ntohl(hdr->reloc_start) + - MAX_SHARED_LIBS * sizeof(unsigned long)); + datapos = ALIGN(realdatastart + + MAX_SHARED_LIBS * sizeof(unsigned long), + FLAT_DATA_ALIGN); + + reloc = (unsigned long *) + (datapos + (ntohl(hdr->reloc_start) - text_len)); memp = textpos; memp_size = len; #ifdef CONFIG_BINFMT_ZFLAT @@ -854,7 +870,7 @@ static int load_flat_binary(struct linux_binprm * bprm, struct pt_regs * regs) stack_len = TOP_OF_ARGS - bprm->p; /* the strings */ stack_len += (bprm->argc + 1) * sizeof(char *); /* the argv array */ stack_len += (bprm->envc + 1) * sizeof(char *); /* the envp array */ - + stack_len += FLAT_DATA_ALIGN - 1; /* reserve for upcoming alignment */ res = load_flat_file(bprm, &libinfo, 0, &stack_len); if (res > (unsigned long)-4096) diff --git a/include/asm-m32r/flat.h b/include/asm-m32r/flat.h index d851cf0..5d711c4 100644 --- a/include/asm-m32r/flat.h +++ b/include/asm-m32r/flat.h @@ -12,7 +12,6 @@ #ifndef __ASM_M32R_FLAT_H #define __ASM_M32R_FLAT_H -#define flat_stack_align(sp) (*sp += (*sp & 3 ? (4 - (*sp & 3)): 0)) #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_set_persistent(relval, p) 0 ^ permalink raw reply related [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-03-05 17:00 [PATCH -v3] flat: fix data sections alignment Oskar Schirmer @ 2009-03-10 7:33 ` Paul Mundt 2009-04-02 20:53 ` Andrew Morton 0 siblings, 1 reply; 14+ messages in thread From: Paul Mundt @ 2009-03-10 7:33 UTC (permalink / raw) To: Oskar Schirmer Cc: Andrew Morton, Johannes Weiner, David Howells, Russell King, Bryan Wu, Geert Uytterhoeven, Mike Frysinger, Greg Ungerer, linux-kernel On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: > The flat loader uses an architecture's flat_stack_align() to align the > stack but assumes word-alignment is enough for the data sections. > > However, on the Xtensa S6000 we have registers up to 128bit width > which can be used from userspace and therefor need userspace stack and > data-section alignment of at least this size. > > This patch drops flat_stack_align() and uses the same alignment that > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's > not defined by the architecture. > > It also fixes m32r which was obviously kaput, aligning an > uninitialized stack entry instead of the stack pointer. > > Signed-off-by: Oskar Schirmer <os@emlix.com> > Signed-off-by: Johannes Weiner <jw@emlix.com> > Cc: David Howells <dhowells@redhat.com> > Cc: Russell King <rmk@arm.linux.org.uk> > Cc: Bryan Wu <cooloney@kernel.org> > Cc: Geert Uytterhoeven <geert@linux-m68k.org> > Cc: Paul Mundt <lethal@linux-sh.org> > Cc: Greg Ungerer <gerg@uclinux.org> > Cc: Mike Frysinger <vapier.adi@gmail.com> The updated version looks ok to me anyways, and it's certainly an improvement over defining the same alignment requirements all over the place. Acked-by: Paul Mundt <lethal@linux-sh.org> ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-03-10 7:33 ` Paul Mundt @ 2009-04-02 20:53 ` Andrew Morton 2009-04-09 14:25 ` Johannes Weiner 0 siblings, 1 reply; 14+ messages in thread From: Andrew Morton @ 2009-04-02 20:53 UTC (permalink / raw) To: Paul Mundt Cc: os, jw, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Tue, 10 Mar 2009 16:33:50 +0900 Paul Mundt <lethal@linux-sh.org> wrote: > On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: > > The flat loader uses an architecture's flat_stack_align() to align the > > stack but assumes word-alignment is enough for the data sections. > > > > However, on the Xtensa S6000 we have registers up to 128bit width > > which can be used from userspace and therefor need userspace stack and > > data-section alignment of at least this size. > > > > This patch drops flat_stack_align() and uses the same alignment that > > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's > > not defined by the architecture. > > > > It also fixes m32r which was obviously kaput, aligning an > > uninitialized stack entry instead of the stack pointer. > > > > Signed-off-by: Oskar Schirmer <os@emlix.com> > > Signed-off-by: Johannes Weiner <jw@emlix.com> > > Cc: David Howells <dhowells@redhat.com> > > Cc: Russell King <rmk@arm.linux.org.uk> > > Cc: Bryan Wu <cooloney@kernel.org> > > Cc: Geert Uytterhoeven <geert@linux-m68k.org> > > Cc: Paul Mundt <lethal@linux-sh.org> > > Cc: Greg Ungerer <gerg@uclinux.org> > > Cc: Mike Frysinger <vapier.adi@gmail.com> > > The updated version looks ok to me anyways, and it's certainly an > improvement over defining the same alignment requirements all over the > place. > > Acked-by: Paul Mundt <lethal@linux-sh.org> Quite a few people expressed quite a few reservations over v2. Are we all OK with a v3 merge? From: Oskar Schirmer <os@emlix.com> The flat loader uses an architecture's flat_stack_align() to align the stack but assumes word-alignment is enough for the data sections. However, on the Xtensa S6000 we have registers up to 128bit width which can be used from userspace and therefor need userspace stack and data-section alignment of at least this size. This patch drops flat_stack_align() and uses the same alignment that is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's not defined by the architecture. It also fixes m32r which was obviously kaput, aligning an uninitialized stack entry instead of the stack pointer. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Oskar Schirmer <os@emlix.com> Cc: David Howells <dhowells@redhat.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Bryan Wu <cooloney@kernel.org> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Acked-by: Paul Mundt <lethal@linux-sh.org> Cc: Greg Ungerer <gerg@uclinux.org> Signed-off-by: Johannes Weiner <jw@emlix.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> --- arch/arm/include/asm/flat.h | 3 - arch/blackfin/include/asm/flat.h | 1 arch/h8300/include/asm/flat.h | 1 arch/m68k/include/asm/flat.h | 1 arch/sh/include/asm/flat.h | 1 fs/binfmt_flat.c | 46 +++++++++++++++++++---------- include/asm-m32r/flat.h | 1 7 files changed, 31 insertions(+), 23 deletions(-) diff -puN arch/arm/include/asm/flat.h~flat-fix-data-sections-alignment arch/arm/include/asm/flat.h --- a/arch/arm/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/arm/include/asm/flat.h @@ -5,9 +5,6 @@ #ifndef __ARM_FLAT_H__ #define __ARM_FLAT_H__ -/* An odd number of words will be pushed after this alignment, so - deliberately misalign the value. */ -#define flat_stack_align(sp) sp = (void *)(((unsigned long)(sp) - 4) | 4) #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN arch/blackfin/include/asm/flat.h~flat-fix-data-sections-alignment arch/blackfin/include/asm/flat.h --- a/arch/blackfin/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/blackfin/include/asm/flat.h @@ -10,7 +10,6 @@ #include <asm/unaligned.h> -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) diff -puN arch/h8300/include/asm/flat.h~flat-fix-data-sections-alignment arch/h8300/include/asm/flat.h --- a/arch/h8300/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/h8300/include/asm/flat.h @@ -5,7 +5,6 @@ #ifndef __H8300_FLAT_H__ #define __H8300_FLAT_H__ -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) 1 #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN arch/m68k/include/asm/flat.h~flat-fix-data-sections-alignment arch/m68k/include/asm/flat.h --- a/arch/m68k/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/m68k/include/asm/flat.h @@ -5,7 +5,6 @@ #ifndef __M68KNOMMU_FLAT_H__ #define __M68KNOMMU_FLAT_H__ -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN arch/sh/include/asm/flat.h~flat-fix-data-sections-alignment arch/sh/include/asm/flat.h --- a/arch/sh/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/sh/include/asm/flat.h @@ -12,7 +12,6 @@ #ifndef __ASM_SH_FLAT_H #define __ASM_SH_FLAT_H -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN fs/binfmt_flat.c~flat-fix-data-sections-alignment fs/binfmt_flat.c --- a/fs/binfmt_flat.c~flat-fix-data-sections-alignment +++ a/fs/binfmt_flat.c @@ -41,6 +41,7 @@ #include <asm/uaccess.h> #include <asm/unaligned.h> #include <asm/cacheflush.h> +#include <asm/page.h> /****************************************************************************/ @@ -54,6 +55,18 @@ #define DBG_FLT(a...) #endif +/* + * User data (stack, data section and bss) needs to be aligned + * for the same reasons as SLAB memory is, and to the same amount. + * Avoid duplicating architecture specific code by using the same + * macro as with SLAB allocation: + */ +#ifdef ARCH_SLAB_MINALIGN +#define FLAT_DATA_ALIGN (ARCH_SLAB_MINALIGN) +#else +#define FLAT_DATA_ALIGN (sizeof(void *)) +#endif + #define RELOC_FAILED 0xff00ff01 /* Relocation incorrect somewhere */ #define UNLOADED_LIB 0x7ff000ff /* Placeholder for unused library */ @@ -114,20 +127,18 @@ static unsigned long create_flat_tables( int envc = bprm->envc; char uninitialized_var(dummy); - sp = (unsigned long *) ((-(unsigned long)sizeof(char *))&(unsigned long) p); + sp = (unsigned long *)p; + sp -= (envc + argc + 2) + 1 + (flat_argvp_envp_on_stack() ? 2 : 0); + sp = (unsigned long *) ((unsigned long)sp & -FLAT_DATA_ALIGN); + argv = sp + 1 + (flat_argvp_envp_on_stack() ? 2 : 0); + envp = argv + (argc + 1); - sp -= envc+1; - envp = sp; - sp -= argc+1; - argv = sp; - - flat_stack_align(sp); if (flat_argvp_envp_on_stack()) { - --sp; put_user((unsigned long) envp, sp); - --sp; put_user((unsigned long) argv, sp); + put_user((unsigned long) envp, sp + 2); + put_user((unsigned long) argv, sp + 1); } - put_user(argc,--sp); + put_user(argc, sp); current->mm->arg_start = (unsigned long) p; while (argc-->0) { put_user((unsigned long) p, argv++); @@ -558,7 +569,9 @@ static int load_flat_file(struct linux_b ret = realdatastart; goto err; } - datapos = realdatastart + MAX_SHARED_LIBS * sizeof(unsigned long); + datapos = ALIGN(realdatastart + + MAX_SHARED_LIBS * sizeof(unsigned long), + FLAT_DATA_ALIGN); DBG_FLT("BINFMT_FLAT: Allocated data+bss+stack (%d bytes): %x\n", (int)(data_len + bss_len + stack_len), (int)datapos); @@ -604,9 +617,12 @@ static int load_flat_file(struct linux_b } realdatastart = textpos + ntohl(hdr->data_start); - datapos = realdatastart + MAX_SHARED_LIBS * sizeof(unsigned long); - reloc = (unsigned long *) (textpos + ntohl(hdr->reloc_start) + - MAX_SHARED_LIBS * sizeof(unsigned long)); + datapos = ALIGN(realdatastart + + MAX_SHARED_LIBS * sizeof(unsigned long), + FLAT_DATA_ALIGN); + + reloc = (unsigned long *) + (datapos + (ntohl(hdr->reloc_start) - text_len)); memp = textpos; memp_size = len; #ifdef CONFIG_BINFMT_ZFLAT @@ -854,7 +870,7 @@ static int load_flat_binary(struct linux stack_len = TOP_OF_ARGS - bprm->p; /* the strings */ stack_len += (bprm->argc + 1) * sizeof(char *); /* the argv array */ stack_len += (bprm->envc + 1) * sizeof(char *); /* the envp array */ - + stack_len += FLAT_DATA_ALIGN - 1; /* reserve for upcoming alignment */ res = load_flat_file(bprm, &libinfo, 0, &stack_len); if (res > (unsigned long)-4096) diff -puN include/asm-m32r/flat.h~flat-fix-data-sections-alignment include/asm-m32r/flat.h --- a/include/asm-m32r/flat.h~flat-fix-data-sections-alignment +++ a/include/asm-m32r/flat.h @@ -12,7 +12,6 @@ #ifndef __ASM_M32R_FLAT_H #define __ASM_M32R_FLAT_H -#define flat_stack_align(sp) (*sp += (*sp & 3 ? (4 - (*sp & 3)): 0)) #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_set_persistent(relval, p) 0 _ ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-04-02 20:53 ` Andrew Morton @ 2009-04-09 14:25 ` Johannes Weiner 2009-05-05 14:08 ` Johannes Weiner 2009-05-27 18:05 ` Mike Frysinger 0 siblings, 2 replies; 14+ messages in thread From: Johannes Weiner @ 2009-04-09 14:25 UTC (permalink / raw) To: Andrew Morton Cc: Paul Mundt, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > On Tue, 10 Mar 2009 16:33:50 +0900 > Paul Mundt <lethal@linux-sh.org> wrote: > > > On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: > > > The flat loader uses an architecture's flat_stack_align() to align the > > > stack but assumes word-alignment is enough for the data sections. > > > > > > However, on the Xtensa S6000 we have registers up to 128bit width > > > which can be used from userspace and therefor need userspace stack and > > > data-section alignment of at least this size. > > > > > > This patch drops flat_stack_align() and uses the same alignment that > > > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's > > > not defined by the architecture. > > > > > > It also fixes m32r which was obviously kaput, aligning an > > > uninitialized stack entry instead of the stack pointer. > > > > > > Signed-off-by: Oskar Schirmer <os@emlix.com> > > > Signed-off-by: Johannes Weiner <jw@emlix.com> > > > Cc: David Howells <dhowells@redhat.com> > > > Cc: Russell King <rmk@arm.linux.org.uk> > > > Cc: Bryan Wu <cooloney@kernel.org> > > > Cc: Geert Uytterhoeven <geert@linux-m68k.org> > > > Cc: Paul Mundt <lethal@linux-sh.org> > > > Cc: Greg Ungerer <gerg@uclinux.org> > > > Cc: Mike Frysinger <vapier.adi@gmail.com> > > > > The updated version looks ok to me anyways, and it's certainly an > > improvement over defining the same alignment requirements all over the > > place. > > > > Acked-by: Paul Mundt <lethal@linux-sh.org> > > Quite a few people expressed quite a few reservations over v2. > > Are we all OK with a v3 merge? Paul and Mike had complaints. Paul acked the last revision. Mike? We need this for upstream to compile on our configuration. Hannes ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-04-09 14:25 ` Johannes Weiner @ 2009-05-05 14:08 ` Johannes Weiner 2009-05-27 11:49 ` Johannes Weiner 2009-05-27 18:05 ` Mike Frysinger 1 sibling, 1 reply; 14+ messages in thread From: Johannes Weiner @ 2009-05-05 14:08 UTC (permalink / raw) To: Andrew Morton Cc: Paul Mundt, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > > On Tue, 10 Mar 2009 16:33:50 +0900 > > Paul Mundt <lethal@linux-sh.org> wrote: > > > > > On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: > > > > The flat loader uses an architecture's flat_stack_align() to align the > > > > stack but assumes word-alignment is enough for the data sections. > > > > > > > > However, on the Xtensa S6000 we have registers up to 128bit width > > > > which can be used from userspace and therefor need userspace stack and > > > > data-section alignment of at least this size. > > > > > > > > This patch drops flat_stack_align() and uses the same alignment that > > > > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's > > > > not defined by the architecture. > > > > > > > > It also fixes m32r which was obviously kaput, aligning an > > > > uninitialized stack entry instead of the stack pointer. > > > > > > > > Signed-off-by: Oskar Schirmer <os@emlix.com> > > > > Signed-off-by: Johannes Weiner <jw@emlix.com> > > > > Cc: David Howells <dhowells@redhat.com> > > > > Cc: Russell King <rmk@arm.linux.org.uk> > > > > Cc: Bryan Wu <cooloney@kernel.org> > > > > Cc: Geert Uytterhoeven <geert@linux-m68k.org> > > > > Cc: Paul Mundt <lethal@linux-sh.org> > > > > Cc: Greg Ungerer <gerg@uclinux.org> > > > > Cc: Mike Frysinger <vapier.adi@gmail.com> > > > > > > The updated version looks ok to me anyways, and it's certainly an > > > improvement over defining the same alignment requirements all over the > > > place. > > > > > > Acked-by: Paul Mundt <lethal@linux-sh.org> > > > > Quite a few people expressed quite a few reservations over v2. > > > > Are we all OK with a v3 merge? > > Paul and Mike had complaints. Paul acked the last revision. Mike? > We need this for upstream to compile on our configuration. Can we get this merged? Nobody complained. Hannes ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-05-05 14:08 ` Johannes Weiner @ 2009-05-27 11:49 ` Johannes Weiner 2009-05-27 20:43 ` Andrew Morton 0 siblings, 1 reply; 14+ messages in thread From: Johannes Weiner @ 2009-05-27 11:49 UTC (permalink / raw) To: Andrew Morton Cc: Paul Mundt, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Tue, May 05, 2009 at 04:08:47PM +0200, Johannes Weiner wrote: > On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: > > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > > > On Tue, 10 Mar 2009 16:33:50 +0900 > > > Paul Mundt <lethal@linux-sh.org> wrote: > > > > > > > On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: > > > > > The flat loader uses an architecture's flat_stack_align() to align the > > > > > stack but assumes word-alignment is enough for the data sections. > > > > > > > > > > However, on the Xtensa S6000 we have registers up to 128bit width > > > > > which can be used from userspace and therefor need userspace stack and > > > > > data-section alignment of at least this size. > > > > > > > > > > This patch drops flat_stack_align() and uses the same alignment that > > > > > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's > > > > > not defined by the architecture. > > > > > > > > > > It also fixes m32r which was obviously kaput, aligning an > > > > > uninitialized stack entry instead of the stack pointer. > > > > > > > > > > Signed-off-by: Oskar Schirmer <os@emlix.com> > > > > > Signed-off-by: Johannes Weiner <jw@emlix.com> > > > > > Cc: David Howells <dhowells@redhat.com> > > > > > Cc: Russell King <rmk@arm.linux.org.uk> > > > > > Cc: Bryan Wu <cooloney@kernel.org> > > > > > Cc: Geert Uytterhoeven <geert@linux-m68k.org> > > > > > Cc: Paul Mundt <lethal@linux-sh.org> > > > > > Cc: Greg Ungerer <gerg@uclinux.org> > > > > > Cc: Mike Frysinger <vapier.adi@gmail.com> > > > > > > > > The updated version looks ok to me anyways, and it's certainly an > > > > improvement over defining the same alignment requirements all over the > > > > place. > > > > > > > > Acked-by: Paul Mundt <lethal@linux-sh.org> > > > > > > Quite a few people expressed quite a few reservations over v2. > > > > > > Are we all OK with a v3 merge? > > > > Paul and Mike had complaints. Paul acked the last revision. Mike? > > We need this for upstream to compile on our configuration. > > Can we get this merged? Nobody complained. <periodic echo-request> What are the plans on this now? I assume it drowned in other changes... Is it too late to get this in .30 now? ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-05-27 11:49 ` Johannes Weiner @ 2009-05-27 20:43 ` Andrew Morton 2009-05-28 0:08 ` Paul Mundt 0 siblings, 1 reply; 14+ messages in thread From: Andrew Morton @ 2009-05-27 20:43 UTC (permalink / raw) To: Johannes Weiner Cc: lethal, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Wed, 27 May 2009 13:49:56 +0200 Johannes Weiner <jw@emlix.com> wrote: > On Tue, May 05, 2009 at 04:08:47PM +0200, Johannes Weiner wrote: > > On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: > > > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > > > > On Tue, 10 Mar 2009 16:33:50 +0900 > > > > Paul Mundt <lethal@linux-sh.org> wrote: > > > > > > > > > On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: > > > > > > The flat loader uses an architecture's flat_stack_align() to align the > > > > > > stack but assumes word-alignment is enough for the data sections. > > > > > > > > > > > > However, on the Xtensa S6000 we have registers up to 128bit width > > > > > > which can be used from userspace and therefor need userspace stack and > > > > > > data-section alignment of at least this size. > > > > > > > > > > > > This patch drops flat_stack_align() and uses the same alignment that > > > > > > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's > > > > > > not defined by the architecture. > > > > > > > > > > > > It also fixes m32r which was obviously kaput, aligning an > > > > > > uninitialized stack entry instead of the stack pointer. > > > > > > > > > > > > Signed-off-by: Oskar Schirmer <os@emlix.com> > > > > > > Signed-off-by: Johannes Weiner <jw@emlix.com> > > > > > > Cc: David Howells <dhowells@redhat.com> > > > > > > Cc: Russell King <rmk@arm.linux.org.uk> > > > > > > Cc: Bryan Wu <cooloney@kernel.org> > > > > > > Cc: Geert Uytterhoeven <geert@linux-m68k.org> > > > > > > Cc: Paul Mundt <lethal@linux-sh.org> > > > > > > Cc: Greg Ungerer <gerg@uclinux.org> > > > > > > Cc: Mike Frysinger <vapier.adi@gmail.com> > > > > > > > > > > The updated version looks ok to me anyways, and it's certainly an > > > > > improvement over defining the same alignment requirements all over the > > > > > place. > > > > > > > > > > Acked-by: Paul Mundt <lethal@linux-sh.org> > > > > > > > > Quite a few people expressed quite a few reservations over v2. > > > > > > > > Are we all OK with a v3 merge? > > > > > > Paul and Mike had complaints. Paul acked the last revision. Mike? > > > We need this for upstream to compile on our configuration. Who is Mike? (A pox upon email clients which strip real names) > > Can we get this merged? Nobody complained. > > <periodic echo-request> > > What are the plans on this now? I assume it drowned in other changes... > Is it too late to get this in .30 now? The patch had a fairly shaky start and it's been unclear to me that everyone is happy with it and has tested it. arch/arm/include/asm/flat.h | 3 -- arch/blackfin/include/asm/flat.h | 1 arch/h8300/include/asm/flat.h | 1 arch/m32r/include/asm/flat.h | 1 arch/m68k/include/asm/flat.h | 1 arch/sh/include/asm/flat.h | 1 fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- I guess ARM is the major architecture amongst that lot. Have Russell and co had a decent look through this? From: Oskar Schirmer <os@emlix.com> The flat loader uses an architecture's flat_stack_align() to align the stack but assumes word-alignment is enough for the data sections. However, on the Xtensa S6000 we have registers up to 128bit width which can be used from userspace and therefor need userspace stack and data-section alignment of at least this size. This patch drops flat_stack_align() and uses the same alignment that is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's not defined by the architecture. It also fixes m32r which was obviously kaput, aligning an uninitialized stack entry instead of the stack pointer. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Oskar Schirmer <os@emlix.com> Cc: David Howells <dhowells@redhat.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Bryan Wu <cooloney@kernel.org> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Acked-by: Paul Mundt <lethal@linux-sh.org> Cc: Greg Ungerer <gerg@uclinux.org> Signed-off-by: Johannes Weiner <jw@emlix.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> --- arch/arm/include/asm/flat.h | 3 - arch/blackfin/include/asm/flat.h | 1 arch/h8300/include/asm/flat.h | 1 arch/m32r/include/asm/flat.h | 1 arch/m68k/include/asm/flat.h | 1 arch/sh/include/asm/flat.h | 1 fs/binfmt_flat.c | 46 +++++++++++++++++++---------- 7 files changed, 31 insertions(+), 23 deletions(-) diff -puN arch/arm/include/asm/flat.h~flat-fix-data-sections-alignment arch/arm/include/asm/flat.h --- a/arch/arm/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/arm/include/asm/flat.h @@ -5,9 +5,6 @@ #ifndef __ARM_FLAT_H__ #define __ARM_FLAT_H__ -/* An odd number of words will be pushed after this alignment, so - deliberately misalign the value. */ -#define flat_stack_align(sp) sp = (void *)(((unsigned long)(sp) - 4) | 4) #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN arch/blackfin/include/asm/flat.h~flat-fix-data-sections-alignment arch/blackfin/include/asm/flat.h --- a/arch/blackfin/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/blackfin/include/asm/flat.h @@ -10,7 +10,6 @@ #include <asm/unaligned.h> -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) diff -puN arch/h8300/include/asm/flat.h~flat-fix-data-sections-alignment arch/h8300/include/asm/flat.h --- a/arch/h8300/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/h8300/include/asm/flat.h @@ -5,7 +5,6 @@ #ifndef __H8300_FLAT_H__ #define __H8300_FLAT_H__ -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) 1 #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN arch/m32r/include/asm/flat.h~flat-fix-data-sections-alignment arch/m32r/include/asm/flat.h --- a/arch/m32r/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/m32r/include/asm/flat.h @@ -12,7 +12,6 @@ #ifndef __ASM_M32R_FLAT_H #define __ASM_M32R_FLAT_H -#define flat_stack_align(sp) (*sp += (*sp & 3 ? (4 - (*sp & 3)): 0)) #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_set_persistent(relval, p) 0 diff -puN arch/m68k/include/asm/flat.h~flat-fix-data-sections-alignment arch/m68k/include/asm/flat.h --- a/arch/m68k/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/m68k/include/asm/flat.h @@ -5,7 +5,6 @@ #ifndef __M68KNOMMU_FLAT_H__ #define __M68KNOMMU_FLAT_H__ -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 1 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN arch/sh/include/asm/flat.h~flat-fix-data-sections-alignment arch/sh/include/asm/flat.h --- a/arch/sh/include/asm/flat.h~flat-fix-data-sections-alignment +++ a/arch/sh/include/asm/flat.h @@ -12,7 +12,6 @@ #ifndef __ASM_SH_FLAT_H #define __ASM_SH_FLAT_H -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) diff -puN fs/binfmt_flat.c~flat-fix-data-sections-alignment fs/binfmt_flat.c --- a/fs/binfmt_flat.c~flat-fix-data-sections-alignment +++ a/fs/binfmt_flat.c @@ -41,6 +41,7 @@ #include <asm/uaccess.h> #include <asm/unaligned.h> #include <asm/cacheflush.h> +#include <asm/page.h> /****************************************************************************/ @@ -54,6 +55,18 @@ #define DBG_FLT(a...) #endif +/* + * User data (stack, data section and bss) needs to be aligned + * for the same reasons as SLAB memory is, and to the same amount. + * Avoid duplicating architecture specific code by using the same + * macro as with SLAB allocation: + */ +#ifdef ARCH_SLAB_MINALIGN +#define FLAT_DATA_ALIGN (ARCH_SLAB_MINALIGN) +#else +#define FLAT_DATA_ALIGN (sizeof(void *)) +#endif + #define RELOC_FAILED 0xff00ff01 /* Relocation incorrect somewhere */ #define UNLOADED_LIB 0x7ff000ff /* Placeholder for unused library */ @@ -114,20 +127,18 @@ static unsigned long create_flat_tables( int envc = bprm->envc; char uninitialized_var(dummy); - sp = (unsigned long *) ((-(unsigned long)sizeof(char *))&(unsigned long) p); + sp = (unsigned long *)p; + sp -= (envc + argc + 2) + 1 + (flat_argvp_envp_on_stack() ? 2 : 0); + sp = (unsigned long *) ((unsigned long)sp & -FLAT_DATA_ALIGN); + argv = sp + 1 + (flat_argvp_envp_on_stack() ? 2 : 0); + envp = argv + (argc + 1); - sp -= envc+1; - envp = sp; - sp -= argc+1; - argv = sp; - - flat_stack_align(sp); if (flat_argvp_envp_on_stack()) { - --sp; put_user((unsigned long) envp, sp); - --sp; put_user((unsigned long) argv, sp); + put_user((unsigned long) envp, sp + 2); + put_user((unsigned long) argv, sp + 1); } - put_user(argc,--sp); + put_user(argc, sp); current->mm->arg_start = (unsigned long) p; while (argc-->0) { put_user((unsigned long) p, argv++); @@ -558,7 +569,9 @@ static int load_flat_file(struct linux_b ret = realdatastart; goto err; } - datapos = realdatastart + MAX_SHARED_LIBS * sizeof(unsigned long); + datapos = ALIGN(realdatastart + + MAX_SHARED_LIBS * sizeof(unsigned long), + FLAT_DATA_ALIGN); DBG_FLT("BINFMT_FLAT: Allocated data+bss+stack (%d bytes): %x\n", (int)(data_len + bss_len + stack_len), (int)datapos); @@ -604,9 +617,12 @@ static int load_flat_file(struct linux_b } realdatastart = textpos + ntohl(hdr->data_start); - datapos = realdatastart + MAX_SHARED_LIBS * sizeof(unsigned long); - reloc = (unsigned long *) (textpos + ntohl(hdr->reloc_start) + - MAX_SHARED_LIBS * sizeof(unsigned long)); + datapos = ALIGN(realdatastart + + MAX_SHARED_LIBS * sizeof(unsigned long), + FLAT_DATA_ALIGN); + + reloc = (unsigned long *) + (datapos + (ntohl(hdr->reloc_start) - text_len)); memp = textpos; memp_size = len; #ifdef CONFIG_BINFMT_ZFLAT @@ -854,7 +870,7 @@ static int load_flat_binary(struct linux stack_len = TOP_OF_ARGS - bprm->p; /* the strings */ stack_len += (bprm->argc + 1) * sizeof(char *); /* the argv array */ stack_len += (bprm->envc + 1) * sizeof(char *); /* the envp array */ - + stack_len += FLAT_DATA_ALIGN - 1; /* reserve for upcoming alignment */ res = load_flat_file(bprm, &libinfo, 0, &stack_len); if (res > (unsigned long)-4096) _ ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-05-27 20:43 ` Andrew Morton @ 2009-05-28 0:08 ` Paul Mundt 2009-05-28 0:16 ` Andrew Morton 0 siblings, 1 reply; 14+ messages in thread From: Paul Mundt @ 2009-05-28 0:08 UTC (permalink / raw) To: Andrew Morton Cc: Johannes Weiner, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Wed, May 27, 2009 at 01:43:57PM -0700, Andrew Morton wrote: > On Wed, 27 May 2009 13:49:56 +0200 > Johannes Weiner <jw@emlix.com> wrote: > > On Tue, May 05, 2009 at 04:08:47PM +0200, Johannes Weiner wrote: > > > On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: > > > > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > > > > Paul and Mike had complaints. Paul acked the last revision. Mike? > > > > We need this for upstream to compile on our configuration. > > Who is Mike? > > (A pox upon email clients which strip real names) > > > > Can we get this merged? Nobody complained. > > > > <periodic echo-request> > > > > What are the plans on this now? I assume it drowned in other changes... > > Is it too late to get this in .30 now? > > The patch had a fairly shaky start and it's been unclear to me that > everyone is happy with it and has tested it. > > arch/arm/include/asm/flat.h | 3 -- > arch/blackfin/include/asm/flat.h | 1 > arch/h8300/include/asm/flat.h | 1 > arch/m32r/include/asm/flat.h | 1 > arch/m68k/include/asm/flat.h | 1 > arch/sh/include/asm/flat.h | 1 > fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- > > I guess ARM is the major architecture amongst that lot. Have Russell > and co had a decent look through this? > The problem here is that the ARM nommu people seem to have disappeared shortly after getting their changes merged. I see no activity within the kernel for ARM nommu since 2006. Given that xtensa has a need for this in its nommu port and people are working on it presently, there certainly has to be a limit to how far we go in allowing an outdated and effectively abandoned port from blocking progress. Perhaps Russell has a more optimistic outlook on the future of ARM nommu, though. The only architectures in that list presently active in the linux nommu space in terms of development, support, and shipping products, are blackfin, m68k, and sh. This too has been the case for years. Microblaze has recently been added (which will also need the same change to its flat.h), and it seems the xtensa folks are at least presently willing to do more than the bare minimum to keep their port limping along. ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-05-28 0:08 ` Paul Mundt @ 2009-05-28 0:16 ` Andrew Morton 2009-06-01 7:44 ` Michal Simek 2009-06-02 14:15 ` Oskar Schirmer 0 siblings, 2 replies; 14+ messages in thread From: Andrew Morton @ 2009-05-28 0:16 UTC (permalink / raw) To: Paul Mundt Cc: jw, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Thu, 28 May 2009 09:08:02 +0900 Paul Mundt <lethal@linux-sh.org> wrote: > On Wed, May 27, 2009 at 01:43:57PM -0700, Andrew Morton wrote: > > On Wed, 27 May 2009 13:49:56 +0200 > > Johannes Weiner <jw@emlix.com> wrote: > > > On Tue, May 05, 2009 at 04:08:47PM +0200, Johannes Weiner wrote: > > > > On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: > > > > > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > > > > > Paul and Mike had complaints. Paul acked the last revision. Mike? > > > > > We need this for upstream to compile on our configuration. > > > > Who is Mike? > > > > (A pox upon email clients which strip real names) > > > > > > Can we get this merged? Nobody complained. > > > > > > <periodic echo-request> > > > > > > What are the plans on this now? I assume it drowned in other changes... > > > Is it too late to get this in .30 now? > > > > The patch had a fairly shaky start and it's been unclear to me that > > everyone is happy with it and has tested it. > > > > arch/arm/include/asm/flat.h | 3 -- > > arch/blackfin/include/asm/flat.h | 1 > > arch/h8300/include/asm/flat.h | 1 > > arch/m32r/include/asm/flat.h | 1 > > arch/m68k/include/asm/flat.h | 1 > > arch/sh/include/asm/flat.h | 1 > > fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- > > > > I guess ARM is the major architecture amongst that lot. Have Russell > > and co had a decent look through this? > > > The problem here is that the ARM nommu people seem to have disappeared > shortly after getting their changes merged. I see no activity within the > kernel for ARM nommu since 2006. heh, OK. Oskar touched it last, so he's now ARM nommu maintainer. > Given that xtensa has a need for this in > its nommu port and people are working on it presently, there certainly > has to be a limit to how far we go in allowing an outdated and > effectively abandoned port from blocking progress. Perhaps Russell has a > more optimistic outlook on the future of ARM nommu, though. > > The only architectures in that list presently active in the linux nommu > space in terms of development, support, and shipping products, are > blackfin, m68k, and sh. This too has been the case for years. Microblaze > has recently been added (which will also need the same change to its > flat.h), and it seems the xtensa folks are at least presently willing to > do more than the bare minimum to keep their port limping along. OK, thanks, I'll send it in to Linus this week. ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-05-28 0:16 ` Andrew Morton @ 2009-06-01 7:44 ` Michal Simek 2009-06-02 14:15 ` Oskar Schirmer 1 sibling, 0 replies; 14+ messages in thread From: Michal Simek @ 2009-06-01 7:44 UTC (permalink / raw) To: Andrew Morton Cc: Paul Mundt, jw, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel Andrew Morton wrote: > On Thu, 28 May 2009 09:08:02 +0900 > Paul Mundt <lethal@linux-sh.org> wrote: > > >> On Wed, May 27, 2009 at 01:43:57PM -0700, Andrew Morton wrote: >> >>> On Wed, 27 May 2009 13:49:56 +0200 >>> Johannes Weiner <jw@emlix.com> wrote: >>> >>>> On Tue, May 05, 2009 at 04:08:47PM +0200, Johannes Weiner wrote: >>>> >>>>> On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: >>>>> >>>>>> On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: >>>>>> Paul and Mike had complaints. Paul acked the last revision. Mike? >>>>>> We need this for upstream to compile on our configuration. >>>>>> >>> Who is Mike? >>> >>> (A pox upon email clients which strip real names) >>> >>> >>>>> Can we get this merged? Nobody complained. >>>>> >>>> <periodic echo-request> >>>> >>>> What are the plans on this now? I assume it drowned in other changes... >>>> Is it too late to get this in .30 now? >>>> >>> The patch had a fairly shaky start and it's been unclear to me that >>> everyone is happy with it and has tested it. >>> >>> arch/arm/include/asm/flat.h | 3 -- >>> arch/blackfin/include/asm/flat.h | 1 >>> arch/h8300/include/asm/flat.h | 1 >>> arch/m32r/include/asm/flat.h | 1 >>> arch/m68k/include/asm/flat.h | 1 >>> arch/sh/include/asm/flat.h | 1 >>> fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- >>> >>> I guess ARM is the major architecture amongst that lot. Have Russell >>> and co had a decent look through this? >>> >>> >> The problem here is that the ARM nommu people seem to have disappeared >> shortly after getting their changes merged. I see no activity within the >> kernel for ARM nommu since 2006. >> > > heh, OK. Oskar touched it last, so he's now ARM nommu maintainer. > > >> Given that xtensa has a need for this in >> its nommu port and people are working on it presently, there certainly >> has to be a limit to how far we go in allowing an outdated and >> effectively abandoned port from blocking progress. Perhaps Russell has a >> more optimistic outlook on the future of ARM nommu, though. >> >> The only architectures in that list presently active in the linux nommu >> space in terms of development, support, and shipping products, are >> blackfin, m68k, and sh. This too has been the case for years. Microblaze >> has recently been added (which will also need the same change to its >> flat.h), and it seems the xtensa folks are at least presently willing to >> do more than the bare minimum to keep their port limping along. >> > > OK, thanks, I'll send it in to Linus this week. > Can you remove flat_stack_align for Microblaze too? It should be in this patch too. Thanks, Michal diff --git a/arch/microblaze/include/asm/flat.h b/arch/microblaze/include/asm/flat.h index acf0da5..6847c15 100644 --- a/arch/microblaze/include/asm/flat.h +++ b/arch/microblaze/include/asm/flat.h @@ -13,7 +13,6 @@ #include <asm/unaligned.h> -#define flat_stack_align(sp) /* nothing needed */ #define flat_argvp_envp_on_stack() 0 #define flat_old_ram_flag(flags) (flags) #define flat_reloc_valid(reloc, size) ((reloc) <= (size)) > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ > -- Michal Simek, Ing. (M.Eng) PetaLogix - Linux Solutions for a Reconfigurable World w: www.petalogix.com p: +61-7-30090663,+42-0-721842854 f: +61-7-30090663 ^ permalink raw reply related [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-05-28 0:16 ` Andrew Morton 2009-06-01 7:44 ` Michal Simek @ 2009-06-02 14:15 ` Oskar Schirmer 2009-06-02 18:57 ` Mike Frysinger 1 sibling, 1 reply; 14+ messages in thread From: Oskar Schirmer @ 2009-06-02 14:15 UTC (permalink / raw) To: Andrew Morton Cc: Paul Mundt, jw, os, dhowells, rmk, cooloney, geert, vapier.adi, gerg, linux-kernel On Wed, May 27, 2009 at 17:16:31 -0700, Andrew Morton wrote: > On Thu, 28 May 2009 09:08:02 +0900 > Paul Mundt <lethal@linux-sh.org> wrote: > > > On Wed, May 27, 2009 at 01:43:57PM -0700, Andrew Morton wrote: > > > On Wed, 27 May 2009 13:49:56 +0200 > > > Johannes Weiner <jw@emlix.com> wrote: > > > > On Tue, May 05, 2009 at 04:08:47PM +0200, Johannes Weiner wrote: > > > > > On Thu, Apr 09, 2009 at 04:25:53PM +0200, Johannes Weiner wrote: > > > > > > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: > > > > > > Paul and Mike had complaints. Paul acked the last revision. Mike? > > > > > > We need this for upstream to compile on our configuration. > > > > > > Who is Mike? > > > > > > (A pox upon email clients which strip real names) > > > > > > > > Can we get this merged? Nobody complained. > > > > > > > > <periodic echo-request> > > > > > > > > What are the plans on this now? I assume it drowned in other changes... > > > > Is it too late to get this in .30 now? > > > > > > The patch had a fairly shaky start and it's been unclear to me that > > > everyone is happy with it and has tested it. > > > > > > arch/arm/include/asm/flat.h | 3 -- > > > arch/blackfin/include/asm/flat.h | 1 > > > arch/h8300/include/asm/flat.h | 1 > > > arch/m32r/include/asm/flat.h | 1 > > > arch/m68k/include/asm/flat.h | 1 > > > arch/sh/include/asm/flat.h | 1 > > > fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- > > > > > > I guess ARM is the major architecture amongst that lot. Have Russell > > > and co had a decent look through this? > > > > > The problem here is that the ARM nommu people seem to have disappeared > > shortly after getting their changes merged. I see no activity within the > > kernel for ARM nommu since 2006. > > heh, OK. Oskar touched it last, so he's now ARM nommu maintainer. Well, I'ld need some test device for that purpose. Someone around who can provide one? Oskar ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-06-02 14:15 ` Oskar Schirmer @ 2009-06-02 18:57 ` Mike Frysinger 2009-06-10 4:34 ` Greg Ungerer 0 siblings, 1 reply; 14+ messages in thread From: Mike Frysinger @ 2009-06-02 18:57 UTC (permalink / raw) To: Oskar Schirmer Cc: Andrew Morton, Paul Mundt, jw, dhowells, rmk, cooloney, geert, gerg, linux-kernel On Tue, Jun 2, 2009 at 10:15, Oskar Schirmer wrote: > On Wed, May 27, 2009 at 17:16:31 -0700, Andrew Morton wrote: >> On Thu, 28 May 2009 09:08:02 +0900 >> Paul Mundt <lethal@linux-sh.org> wrote: >> > On Wed, May 27, 2009 at 01:43:57PM -0700, Andrew Morton wrote: >> > > The patch had a fairly shaky start and it's been unclear to me that >> > > everyone is happy with it and has tested it. >> > > >> > > arch/arm/include/asm/flat.h | 3 -- >> > > arch/blackfin/include/asm/flat.h | 1 >> > > arch/h8300/include/asm/flat.h | 1 >> > > arch/m32r/include/asm/flat.h | 1 >> > > arch/m68k/include/asm/flat.h | 1 >> > > arch/sh/include/asm/flat.h | 1 >> > > fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- >> > > >> > > I guess ARM is the major architecture amongst that lot. Have Russell >> > > and co had a decent look through this? >> > > >> > The problem here is that the ARM nommu people seem to have disappeared >> > shortly after getting their changes merged. I see no activity within the >> > kernel for ARM nommu since 2006. >> >> heh, OK. Oskar touched it last, so he's now ARM nommu maintainer. > > Well, I'ld need some test device for that purpose. > Someone around who can provide one? seems a bit odd for a maintainer to not have any hardware ... at any rate, dont you have an ARM system ? cant you just disable support for the MMU ? -mike ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-06-02 18:57 ` Mike Frysinger @ 2009-06-10 4:34 ` Greg Ungerer 0 siblings, 0 replies; 14+ messages in thread From: Greg Ungerer @ 2009-06-10 4:34 UTC (permalink / raw) To: Mike Frysinger Cc: Oskar Schirmer, Andrew Morton, Paul Mundt, jw, dhowells, rmk, cooloney, geert, linux-kernel Hi Mike, Mike Frysinger wrote: > On Tue, Jun 2, 2009 at 10:15, Oskar Schirmer wrote: >> On Wed, May 27, 2009 at 17:16:31 -0700, Andrew Morton wrote: >>> On Thu, 28 May 2009 09:08:02 +0900 >>> Paul Mundt <lethal@linux-sh.org> wrote: >>>> On Wed, May 27, 2009 at 01:43:57PM -0700, Andrew Morton wrote: >>>>> The patch had a fairly shaky start and it's been unclear to me that >>>>> everyone is happy with it and has tested it. >>>>> >>>>> arch/arm/include/asm/flat.h | 3 -- >>>>> arch/blackfin/include/asm/flat.h | 1 >>>>> arch/h8300/include/asm/flat.h | 1 >>>>> arch/m32r/include/asm/flat.h | 1 >>>>> arch/m68k/include/asm/flat.h | 1 >>>>> arch/sh/include/asm/flat.h | 1 >>>>> fs/binfmt_flat.c | 46 ++++++++++++++++++++++++++------------- >>>>> >>>>> I guess ARM is the major architecture amongst that lot. Have Russell >>>>> and co had a decent look through this? >>>>> >>>> The problem here is that the ARM nommu people seem to have disappeared >>>> shortly after getting their changes merged. I see no activity within the >>>> kernel for ARM nommu since 2006. >>> heh, OK. Oskar touched it last, so he's now ARM nommu maintainer. >> Well, I'ld need some test device for that purpose. >> Someone around who can provide one? > > seems a bit odd for a maintainer to not have any hardware ... at any > rate, dont you have an ARM system ? cant you just disable support for > the MMU ? The nommu support for arm is not complete in mainline. (And that is why CONFIG_MMU is not a settable option in arm yet). Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Principal Engineer EMAIL: gerg@snapgear.com SnapGear Group, McAfee PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH -v3] flat: fix data sections alignment 2009-04-09 14:25 ` Johannes Weiner 2009-05-05 14:08 ` Johannes Weiner @ 2009-05-27 18:05 ` Mike Frysinger 1 sibling, 0 replies; 14+ messages in thread From: Mike Frysinger @ 2009-05-27 18:05 UTC (permalink / raw) To: Johannes Weiner Cc: Andrew Morton, Paul Mundt, os, dhowells, rmk, cooloney, geert, gerg, linux-kernel On Thu, Apr 9, 2009 at 10:25, Johannes Weiner wrote: > On Thu, Apr 02, 2009 at 01:53:43PM -0700, Andrew Morton wrote: >> On Tue, 10 Mar 2009 16:33:50 +0900 >> Paul Mundt <lethal@linux-sh.org> wrote: >> >> > On Thu, Mar 05, 2009 at 06:00:59PM +0100, Oskar Schirmer wrote: >> > > The flat loader uses an architecture's flat_stack_align() to align the >> > > stack but assumes word-alignment is enough for the data sections. >> > > >> > > However, on the Xtensa S6000 we have registers up to 128bit width >> > > which can be used from userspace and therefor need userspace stack and >> > > data-section alignment of at least this size. >> > > >> > > This patch drops flat_stack_align() and uses the same alignment that >> > > is required for slab caches, ARCH_SLAB_MINALIGN, or wordsize if it's >> > > not defined by the architecture. >> > > >> > > It also fixes m32r which was obviously kaput, aligning an >> > > uninitialized stack entry instead of the stack pointer. >> > > >> > > Signed-off-by: Oskar Schirmer <os@emlix.com> >> > > Signed-off-by: Johannes Weiner <jw@emlix.com> >> > > Cc: David Howells <dhowells@redhat.com> >> > > Cc: Russell King <rmk@arm.linux.org.uk> >> > > Cc: Bryan Wu <cooloney@kernel.org> >> > > Cc: Geert Uytterhoeven <geert@linux-m68k.org> >> > > Cc: Paul Mundt <lethal@linux-sh.org> >> > > Cc: Greg Ungerer <gerg@uclinux.org> >> > > Cc: Mike Frysinger <vapier.adi@gmail.com> >> > >> > The updated version looks ok to me anyways, and it's certainly an >> > improvement over defining the same alignment requirements all over the >> > place. >> > >> > Acked-by: Paul Mundt <lethal@linux-sh.org> >> >> Quite a few people expressed quite a few reservations over v2. >> >> Are we all OK with a v3 merge? > > Paul and Mike had complaints. Paul acked the last revision. Mike? > We need this for upstream to compile on our configuration. i thought my comments had all been addressed, so i dont have any problems -mike ^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2009-06-10 4:34 UTC | newest] Thread overview: 14+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2009-03-05 17:00 [PATCH -v3] flat: fix data sections alignment Oskar Schirmer 2009-03-10 7:33 ` Paul Mundt 2009-04-02 20:53 ` Andrew Morton 2009-04-09 14:25 ` Johannes Weiner 2009-05-05 14:08 ` Johannes Weiner 2009-05-27 11:49 ` Johannes Weiner 2009-05-27 20:43 ` Andrew Morton 2009-05-28 0:08 ` Paul Mundt 2009-05-28 0:16 ` Andrew Morton 2009-06-01 7:44 ` Michal Simek 2009-06-02 14:15 ` Oskar Schirmer 2009-06-02 18:57 ` Mike Frysinger 2009-06-10 4:34 ` Greg Ungerer 2009-05-27 18:05 ` Mike Frysinger
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.