summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* sched: Move nr_cpus_allowed out of 'struct sched_rt_entity'Peter Zijlstra2012-05-306-20/+26
| | | | | | | | | Since nr_cpus_allowed is used outside of sched/rt.c and wants to be used outside of there more, move it to a more natural site. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/n/tip-kr61f02y9brwzkh6x53pdptm@git.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
* sched: Make sure to not re-read variables after validationPeter Zijlstra2012-05-301-4/+11
| | | | | | | | | | We could re-read rq->rt_avg after we validated it was smaller than total, invalidating the check and resulting in an unintended negative. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: David Rientjes <rientjes@google.com> Link: http://lkml.kernel.org/r/1337688268.9698.29.camel@twins Signed-off-by: Ingo Molnar <mingo@kernel.org>
* sched: Fix SD_OVERLAPPeter Zijlstra2012-05-302-7/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | SD_OVERLAP exists to allow overlapping groups, overlapping groups appear in NUMA topologies that aren't fully connected. The typical result of not fully connected NUMA is that each cpu (or rather node) will have different spans for a particular distance. However due to how sched domains are traversed -- only the first cpu in the mask goes one level up -- the next level only cares about the spans of the cpus that went up. Due to this two things were observed to be broken: - build_overlap_sched_groups() -- since its possible the cpu we're building the groups for exists in multiple (or all) groups, the selection criteria of the first group didn't ensure there was a cpu for which is was true that cpumask_first(span) == cpu. Thus load- balancing would terminate. - update_group_power() -- assumed that the cpu span of the first group of the domain was covered by all groups of the child domain. The above explains why this isn't true, so deal with it. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: David Rientjes <rientjes@google.com> Link: http://lkml.kernel.org/r/1337788843.9783.14.camel@laptop Signed-off-by: Ingo Molnar <mingo@kernel.org>
* sched: Don't try allocating memory from offline nodesPeter Zijlstra2012-05-301-1/+1
| | | | | | | | | | | Allocators don't appreciate it when you try and allocate memory from offline nodes. Reported-and-tested-by: Tony Luck <tony.luck@intel.com> Reported-and-tested-by: Anton Blanchard <anton@samba.org> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/n/tip-epfc1io9whb7o22bcujf31vn@git.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
* sched/nohz: Fix rq->cpu_load calculations some morePeter Zijlstra2012-05-303-10/+45
| | | | | | | | | | | | | | | | | | | | | | Follow up on commit 556061b00 ("sched/nohz: Fix rq->cpu_load[] calculations") since while that fixed the busy case it regressed the mostly idle case. Add a callback from the nohz exit to also age the rq->cpu_load[] array. This closes the hole where either there was no nohz load balance pass during the nohz, or there was a 'significant' amount of idle time between the last nohz balance and the nohz exit. So we'll update unconditionally from the tick to not insert any accidental 0 load periods while busy, and we try and catch up from nohz idle balance and nohz exit. Both these are still prone to missing a jiffy, but that has always been the case. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: pjt@google.com Cc: Venkatesh Pallipadi <venki@google.com> Link: http://lkml.kernel.org/n/tip-kt0trz0apodbf84ucjfdbr1a@git.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
* sched/x86: Use cpu_llc_shared_mask(cpu) for coregroup_maskPeter Zijlstra2012-05-301-9/+1
| | | | | | | | | | | | | | | Commit commit 8e7fbcbc2 ("sched: Remove stale power aware scheduling remnants and dysfunctional knobs") made a boo-boo with removing the power aware scheduling muck from the x86 topology bits. We should unconditionally use the llc_shared mask for multi-core. Reported-and-tested-by: Mike Galbraith <efault@gmx.de> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Borislav Petkov <bp@amd64.org> Cc: Andreas Herrmann <andreas.herrmann3@amd.com> Link: http://lkml.kernel.org/n/tip-lsksc2kfyeveb13avh327p0d@git.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
* Merge branch 'x86-trampoline-for-linus' of ↵Linus Torvalds2012-05-2959-613/+774
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip Pull x86 trampoline rework from H. Peter Anvin: "This code reworks all the "trampoline"/"realmode" code (various bits that need to live in the first megabyte of memory, most but not all of which runs in real mode at some point) in the kernel into a single object. The main reason for doing this is that it eliminates the last place in the kernel where we needed pages to be mapped RWX. This code separates all that code into proper R/RW/RX pages." Fix up conflicts in arch/x86/kernel/Makefile (mca removed next to reboot code), and arch/x86/kernel/reboot.c (reboot code moved around in one branch, modified in this one), and arch/x86/tools/relocs.c (mostly same code came in earlier due to working around the ld bugs just before the 3.4 release). Also remove stale x86-relocs entry from scripts/.gitignore as per Peter Anvin. * commit '61f5446169046c217a5479517edac3a890c3bee7': (36 commits) x86, realmode: Move end signature into header.S x86, relocs: When printing an error, say relative or absolute x86, relocs: More relocations which may end up as absolute x86, relocs: Workaround for binutils 2.22.52.0.1 section bug xen-acpi-processor: Add missing #include <xen/xen.h> acpi, bgrd: Add missing <linux/io.h> to drivers/acpi/bgrt.c x86, realmode: Change EFER to a single u64 field x86, realmode: Move kernel/realmode.c to realmode/init.c x86, realmode: Move not-common bits out of trampoline_common.S x86, realmode: Mask out EFER.LMA when saving trampoline EFER x86, realmode: Fix no cache bits test in reboot_32.S x86, realmode: Make sure all generated files are listed in targets x86, realmode: build fix: remove duplicate build x86, realmode: read cr4 and EFER from kernel for 64-bit trampoline x86, realmode: fixes compilation issue in tboot.c x86, realmode: move relocs from scripts/ to arch/x86/tools x86, realmode: header for trampoline code x86, realmode: flattened rm hierachy x86, realmode: don't copy real_mode_header x86, realmode: fix 64-bit wakeup sequence ...
| * x86, realmode: Move end signature into header.SH. Peter Anvin2012-05-214-6/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The end signature was defined in wakeup_asm.S as it originally came from the ACPI wakeup code. However, we rely on the existence of the .signature section to expand .bss, otherwise we would have to include code to explicitly zero the .bss depending on the configuration. Since the expanded .bss is just in .init.data anyway, it's easier to always have it expanded. This fixes failures when compiled without CONFIG_ACPI_SLEEP. Reported-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: H. Peter Anvin <hpa@zytor.com> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com>
| * x86, relocs: When printing an error, say relative or absoluteH. Peter Anvin2012-05-181-3/+8
| | | | | | | | | | | | | | | | When the relocs tool throws an error, let the error message say if it is an absolute or relative symbol. This should make it a lot more clear what action the programmer needs to take. Signed-off-by: H. Peter Anvin <hpa@zytor.com>
| * x86, relocs: More relocations which may end up as absoluteH. Peter Anvin2012-05-181-1/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | GNU ld 2.22.52.0.1 has a bug that it blindly changes symbols from section-relative to absolute if they are in a section of zero length. This turns the symbols __init_begin and __init_end into absolute symbols. Let the relocs program know that those should be treated as relative symbols. This bug is exposed by checkin 433de739bbc2 x86, realmode: 16-bit real-mode code support for relocs tool only in the sense that that checkin changes the relocs tool to report an error instead of silently generating a kernel which is broken if relocated. Reported-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: H. Peter Anvin <hpa@zytor.com> Cc: H.J. Lu <hjl.tools@gmail.com> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com>
| * x86, relocs: Workaround for binutils 2.22.52.0.1 section bugH. Peter Anvin2012-05-181-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | GNU ld 2.22.52.0.1 has a bug that it blindly changes symbols from section-relative to absolute if they are in a section of zero length. This turns the symbols __init_begin and __init_end into absolute symbols. Let the relocs program know that those should be treated as relative symbols. Reported-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: H. Peter Anvin <hpa@zytor.com> Cc: H.J. Lu <hjl.tools@gmail.com>
| * xen-acpi-processor: Add missing #include <xen/xen.h>H. Peter Anvin2012-05-171-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | This file depends on <xen/xen.h>, but the dependency was hidden due to: <asm/acpi.h> -> <asm/trampoline.h> -> <asm/io.h> -> <xen/xen.h> With the removal of <asm/trampoline.h>, this exposed the missing Reported-by: Ingo Molnar <mingo@kernel.org> Cc: Len Brown <lenb@kernel.org> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Signed-off-by: H. Peter Anvin <hpa@zytor.com>
| * acpi, bgrd: Add missing <linux/io.h> to drivers/acpi/bgrt.cH. Peter Anvin2012-05-171-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | The ACPI BGRT driver uses ioremap() and memcpy_fromio(), which means it needs to #include <linux/io.h>. [ hpa: This was hidden because <asm/acpi.h> used to include <asm/trampoline.h>, which then included <asm/io.h> ] Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336597442-9399-1-git-send-email-hpa@linux.intel.com Cc: Matthew Garrett <mjg@redhat.com> Cc: Len Brown <lenb@kernel.org>
| * x86, realmode: Change EFER to a single u64 fieldH. Peter Anvin2012-05-163-7/+5
| | | | | | | | | | | | | | | | | | | | | | Change EFER to be a single u64 field instead of two u32 fields; change the order to maintain alignment. Note that on x86-64 cr4 is really also a 64-bit quantity, although we can only set the low 32 bits from the trampoline code since it is still executing in 32-bit mode at that point. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com>
| * x86, realmode: Move kernel/realmode.c to realmode/init.cH. Peter Anvin2012-05-163-1/+1
| | | | | | | | | | | | | | | | Keep all the realmode code together, including initialization (only the rm/ subdirectory is actually built as real-mode code, anyway.) Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com>
| * x86, realmode: Move not-common bits out of trampoline_common.SH. Peter Anvin2012-05-163-35/+33
| | | | | | | | | | | | | | | | | | | | Move the bits that aren't actually common out of trampoline_common.S and into the arch-specific files. Furthermore, make sure the page directory is first in the .bss section for trampoline_64.S in order to not waste an entire page of memory. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com>
| * x86, realmode: Mask out EFER.LMA when saving trampoline EFERH. Peter Anvin2012-05-161-3/+8
| | | | | | | | | | | | | | | | | | | | | | Some AMD processors apparently #GP(0) if EFER.LMA is set in WRMSR, rather than ignoring it. Thus, we need to mask it out. Reported-by: Ingo Molnar <mingo@kernel.org> Tested-by: Borislav Petkov <bp@alien8.de> Cc: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-24-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Fix no cache bits test in reboot_32.SJarkko Sakkinen2012-05-101-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Before the new real-mode code infrastructure %edx was used for testing CD and NW bits with andl in order to decide whether to flush the processor caches or not. The value of cr0 was also stored in %eax, which was later used to set cr0 after masking out lower byte (except TS bit) in order to enter real-mode. In the new real-mode code infrastructure we wanted to keep input parameter in %eax so we are using %edx for both cr0 cases. This has caused regression since andl overwrites the value of %edx. This patch fixes the issue by replacing andl with testl, which is essentially andl without writing result to the register. Special thanks to Paolo Bonzini for noting this and proposing a fix. Reported-and-tested-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336633898-23743-1-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: Make sure all generated files are listed in targetsH. Peter Anvin2012-05-091-7/+13
| | | | | | | | | | | | | | | | | | | | Kbuild expects all generated files to be listed in the targets variable. If it isn't, weird things happen. Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Michal Marek <mmarek@suse.cz> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336595106-21135-1-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: build fix: remove duplicate buildJarkko Sakkinen2012-05-092-5/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Real-mode binary was built twice. This patch fixes the issue by making realmode.relocs as target for realmode.bin. [ hpa: removed the direct dependency on realmode.relocs in arch/x86/realmode/Makefile ] Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336595106-21135-1-git-send-email-jarkko.sakkinen@intel.com Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Michal Marek <mmarek@suse.cz> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: read cr4 and EFER from kernel for 64-bit trampolineJarkko Sakkinen2012-05-087-28/+49
| | | | | | | | | | | | | | | | | | | | This patch changes 64-bit trampoline so that CR4 and EFER are provided by the kernel instead of using fixed values. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-24-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: fixes compilation issue in tboot.cJarkko Sakkinen2012-05-081-1/+1
| | | | | | | | | | | | | | | | Fixed include path of wakeup.h in tboot.c. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-23-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: move relocs from scripts/ to arch/x86/toolsJarkko Sakkinen2012-05-088-7/+17
| | | | | | | | | | | | | | | | | | | | | | | | Moved relocs tool from scripts/ to arch/x86/tools because it is architecture specific script. Added new target archscripts that can be used to build scripts needed building an architecture. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-22-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Michal Marek <mmarek@suse.cz>
| * x86, realmode: header for trampoline codeJarkko Sakkinen2012-05-088-88/+87
| | | | | | | | | | | | | | | | | | | | | | | | Added header for trampoline code that can be used to supply input data to it. This makes interface between real mode code and kernel cleaner and simpler. Replaced two confusing pointers to level4 pgt in trampoline_64.S with a single pointer to the beginning of the page table. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-21-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: flattened rm hierachyJarkko Sakkinen2012-05-0821-52/+22
| | | | | | | | | | | | | | | | | | | | | | Simplified hierarchy under rm directory to a flat directory because it is not anymore really justified to have own directory for wakeup code. It only adds more complexity. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-20-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: don't copy real_mode_headerJarkko Sakkinen2012-05-0810-43/+35
| | | | | | | | | | | | | | | | | | Replaced copying of real_mode_header with a pointer to beginning of RM memory. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-19-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: fix 64-bit wakeup sequenceJarkko Sakkinen2012-05-089-107/+110
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | There were number of issues in wakeup sequence: - Wakeup stack was placed in hardcoded address. - NX bit in EFER was not enabled. - Initialization incorrectly set physical address of secondary_startup_64. - Some alignment issues. This patch fixes these issues and in addition: - Unifies coding conventions in .S files. - Sets alignments of code and data right. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-18-git-send-email-jarkko.sakkinen@intel.com Originally-by: H. Peter Anvin <hpa@linux.intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Cc: Len Brown <len.brown@intel.com> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: Fix always-zero test in reboot_32.SH. Peter Anvin2012-05-081-1/+1
| | | | | | | | | | | | | | | | | | A test instruction is an "and", and an and with zero is always zero. This would cause us to always take the BIOS path, not the APM path, in case anyone actually cares... Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-17-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Move trampoline_*.S early in the link orderH. Peter Anvin2012-05-081-1/+1
| | | | | | | | | | | | | | | | | | | | Move trampoline_*.S earlier in the link order so it ends up being first in the text segment; since the SIPI vector requires 4K alignment it otherwise ends up padding the .text segment with that much completely unnecessarily. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-16-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Replace open-coded ljmpw with a macroH. Peter Anvin2012-05-085-16/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We cannot code an ljmpw to the real-mode segment directly, because gas refuses to assemble an ljmp with a symbolic segment. Instead of open-coding it everywhere, define a macro and use it for this case. This is specifically an ljmpw from a 16-bit segment. This is okay, as one should never enter real mode from a 32-bit segment: if one do, the CPU ends up in a bizarre (and useless) mode sometimes called "unreal mode" where segments behave like real mode but the default address and operand sizes is 32 bits. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-15-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Remove indirect jumps in trampoline_32 and wakeup_asmH. Peter Anvin2012-05-082-12/+18
| | | | | | | | | | | | | | | | | | | | | | | | Remove indirect jumps in trampoline_32.S and the 32-bit part of wakeup_asm.S. There exist systems which are known to do weird things if an SMI comes in right after a mode switch, and the safest way to deal with it is to always follow with a simple absolute far jump. In the 64-bit code we then to a register indirect near jump; follow that pattern for the 32-bit code. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-14-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Remove indirect jumps in trampoline_64.SH. Peter Anvin2012-05-081-13/+2
| | | | | | | | | | | | | | | | | | Remove indirect jumps in trampoline_64.S which are no longer necessary: the realmode code can relocate the absolute jumps correctly from the start. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-13-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Align .data section in trampoline_32.SH. Peter Anvin2012-05-081-1/+1
| | | | | | | | | | | | | | Specify the alignment of the .data section in trampoline_32.S. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-12-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Move bits to the proper sections in trampoline_64.SH. Peter Anvin2012-05-081-4/+6
| | | | | | | | | | | | | | | | | | | | Move various bits to the sections they really belong in in trampoline_64.S. Use GLOBAL() rather than ENTRY() for data objects: ENTRY() should only be used with code and forces alignment to 16 bytes. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-11-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Add .text64 section, make barrier symbols absoluteH. Peter Anvin2012-05-081-3/+8
| | | | | | | | | | | | | | | | | | | | | | | | Add a .text64 section. The purpose of this is to keep 16-, 32- and 64-bit code segregated into separate sections, mainly to keep disassembly sane. Move barrier symbols out of sections to avoid the "symbol in empty section" problem in some versions of GNU ld. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-10-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Allow absolute pa_* symbols in the realmode codeH. Peter Anvin2012-05-081-0/+7
| | | | | | | | | | | | | | | | | | Allow pa_* symbols to be absolute (outside any section) in the realmode linker script. Some versions of GNU ld are known to be unhappy about symbols defined in a section that is otherwise empty. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-9-git-send-email-jarkko.sakkinen@intel.com
| * x86, realmode: Set permission for real mode pagesJarkko Sakkinen2012-05-081-1/+15
| | | | | | | | | | | | | | | | | | | | Set proper permissions for rodata, text and data, removing the realmode trampoline area as a remaining RWX memory mapping in the kernel. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-8-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: Move ACPI wakeup to unified realmode codeJarkko Sakkinen2012-05-0835-565/+111
| | | | | | | | | | | | | | | | | | | | | | | | | | Migrated ACPI wakeup code to the real-mode blob. Code existing in .x86_trampoline can be completely removed. Static descriptor table in wakeup_asm.S is courtesy of H. Peter Anvin. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-7-git-send-email-jarkko.sakkinen@intel.com Cc: Rafael J. Wysocki <rjw@sisk.pl> Cc: Len Brown <len.brown@intel.com> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: Move SMP trampoline to unified realmode codeJarkko Sakkinen2012-05-089-16/+316
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Migrated SMP trampoline code to the real mode blob. SMP trampoline code is not yet removed from .x86_trampoline because it is needed by the wakeup code. [ hpa: always enable compiling startup_32_smp in head_32.S... it is only a few instructions which go into .init on UP builds, and it makes the rest of the code less #ifdef ugly. ] Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-6-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: Move reboot_32.S to unified realmode codeJarkko Sakkinen2012-05-086-65/+52
| | | | | | | | | | | | | | | | | | Migrated reboot_32.S from x86_trampoline to the real-mode blob. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-5-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: Relocator for realmode codeJarkko Sakkinen2012-05-084-0/+108
| | | | | | | | | | | | | | | | | | | | | | Implements relocator for real mode code that is called as part of setup_arch(). Processes segment relocations and linear relocations. Real-mode code is relocated to a free hole below 1 MB. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-4-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: realmode.bin infrastructureJarkko Sakkinen2012-05-087-1/+189
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Create realmode.bin and realmode.relocs files. Piggy pack them into relocatable object that will be included into .init.data section of the main kernel image. The first file includes binary image of the real-mode code. The latter file includes all relocations. The layout of the binary image is specified in realmode.lds.S. The makefile generates pa_ prefixed symbols for each exported global. These are used in 32-bit code and in realmode header to define symbols that need to be relocated. Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-3-git-send-email-jarkko.sakkinen@intel.com Originally-by: H. Peter Anvin <hpa@linux.intel.com> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
| * x86, realmode: 16-bit real-mode code support for relocs toolH. Peter Anvin2012-05-084-63/+185
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | A new option is added to the relocs tool called '--realmode'. This option causes the generation of 16-bit segment relocations and 32-bit linear relocations for the real-mode code. When the real-mode code is moved to the low-memory during kernel initialization, these relocation entries can be used to relocate the code properly. In the assembly code 16-bit segment relocations must be relative to the 'real_mode_seg' absolute symbol. Linear relocations must be relative to a symbol prefixed with 'pa_'. 16-bit segment relocation is used to load cs:ip in 16-bit code. Linear relocations are used in the 32-bit code for relocatable data references. They are declared in the linker script of the real-mode code. The relocs tool is moved to scripts/x86-relocs.c so it will be compiled before building the arch/x86 tree. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com> Link: http://lkml.kernel.org/r/1336501366-28617-2-git-send-email-jarkko.sakkinen@intel.com Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@intel.com> Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
* | Merge git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edacLinus Torvalds2012-05-2932-1468/+1981
|\ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Pull EDAC internal API changes from Mauro Carvalho Chehab: "This changeset is the first part of a series of patches that fixes the EDAC sybsystem. On this set, it changes the Kernel EDAC API in order to properly represent the Intel i3/i5/i7, Xeon 3xxx/5xxx/7xxx, and Intel E5-xxxx memory controllers. The EDAC core used to assume that: - the DRAM chip select pin is directly accessed by the memory controller - when multiple channels are used, they're all filled with the same type of memory. None of the above premises is true on Intel memory controllers since 2002, when RAMBUS and FB-DIMMs were introduced, and Advanced Memory Buffer or by some similar technologies hides the direct access to the DRAM pins. So, the existing drivers for those chipsets had to lie to the EDAC core, in general telling that just one channel is filled. That produces some hard to understand error messages like: EDAC MC0: CE row 3, channel 0, label "DIMM1": 1 Unknown error(s): memory read error on FATAL area : cpu=0 Err=0008:00c2 (ch=2), addr = 0xad1f73480 => socket=0, Channel=0(mask=2), rank=1 The location information there (row3 channel 0) is completely bogus: it has no physical meaning, and are just some random values that the driver uses to talk with the EDAC core. The error actually happened at CPU socket 0, channel 0, slot 1, but this is not reported anywhere, as the EDAC core doesn't know anything about the memory layout. So, only advanced users that know how the EDAC driver works and that tests their systems to see how DIMMs are mapped can actually benefit for such error logs. This patch series fixes the error report logic, in order to allow the EDAC to expose the memory architecture used by them to the EDAC core. So, as the EDAC core now understands how the memory is organized, it can provide an useful report: EDAC MC0: CE memory read error on DIMM1 (channel:0 slot:1 page:0x364b1b offset:0x600 grain:32 syndrome:0x0 - count:1 area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:4) The location of the DIMM where the error happened is reported by "MC0" (cpu socket #0), at "channel:0 slot:1" location, and matches the physical location of the DIMM. There are two remaining issues not covered by this patch series: - The EDAC sysfs API will still report bogus values. So, userspace tools like edac-utils will still use the bogus data; - Add a new tracepoint-based way to get the binary information about the errors. Those are on a second series of patches (also at -next), but will probably miss the train for 3.5, due to the slow review process." Fix up trivial conflict (due to spelling correction of removed code) in drivers/edac/edac_device.c * git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edac: (42 commits) i7core: fix ranks information at the per-channel struct i5000: Fix the fatal error handling i5100_edac: Fix a warning when compiled with 32 bits i82975x_edac: Test nr_pages earlier to save a few CPU cycles e752x_edac: provide more info about how DIMMS/ranks are mapped i5000_edac: Fix the logic that retrieves memory information i5400_edac: improve debug messages to better represent the filled memory edac: Cleanup the logs for i7core and sb edac drivers edac: Initialize the dimm label with the known information edac: Remove the legacy EDAC ABI x38_edac: convert driver to use the new edac ABI tile_edac: convert driver to use the new edac ABI sb_edac: convert driver to use the new edac ABI r82600_edac: convert driver to use the new edac ABI ppc4xx_edac: convert driver to use the new edac ABI pasemi_edac: convert driver to use the new edac ABI mv64x60_edac: convert driver to use the new edac ABI mpc85xx_edac: convert driver to use the new edac ABI i82975x_edac: convert driver to use the new edac ABI i82875p_edac: convert driver to use the new edac ABI ...
| * | i7core: fix ranks information at the per-channel structMauro Carvalho Chehab2012-05-281-13/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | There is a flag at the per-channel struct that indicates if there are any 4R dimm on it. The way the presence of this flag were reported is not ok, as it might give the false idea that the channel were filled with 2R memories: [ 580.588701] EDAC DEBUG: get_dimm_config: Ch1 phy rd1, wr1 (0x063f7431): 2 ranks, UDIMMs [ 580.588704] EDAC DEBUG: get_dimm_config: dimm 0 1024 Mb offset: 0, bank: 8, rank: 1, row: 0x4000, col: 0x400 (in this case, just one 1R memory is filled on channel 1) So, use a better way to represent the per-channel ranks information. After the patch, it will show: [ 2002.233978] EDAC DEBUG: get_dimm_config: Ch0 phy rd0, wr0 (0x063f7431): UDIMMs [ 2002.233982] EDAC DEBUG: get_dimm_config: dimm 0 1024 Mb offset: 0, bank: 8, rank: 1, row: 0x4000, col: 0x400 [ 2002.233988] EDAC DEBUG: get_dimm_config: dimm 1 1024 Mb offset: 4, bank: 8, rank: 1, row: 0x4000, col: 0x400 (in this case, there isn't any 4R memories) Reported-by: Borislav Petkov <borislav.petkov@amd.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
| * | i5000: Fix the fatal error handlingMauro Carvalho Chehab2012-05-281-7/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | The fatal error channel bits point to a single channel, and not to a range of channels. Fix the code to properly report it, instead of printing messages like: kernel: EDAC MC0: INTERNAL ERROR: channel-b out of range (4 >= 4) Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
| * | i5100_edac: Fix a warning when compiled with 32 bitsMauro Carvalho Chehab2012-05-281-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | drivers/edac/i5100_edac.c: In function ‘i5100_init_csrows’: drivers/edac/i5100_edac.c:862:3: warning: format ‘%zd’ expects argument of type ‘signed size_t’, but argument 5 has type ‘long unsigned int’ [-Wformat] Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
| * | i82975x_edac: Test nr_pages earlier to save a few CPU cyclesMauro Carvalho Chehab2012-05-281-6/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Avoid test nr_pages twice, and initializing some data that won't be used. Cleanup patch only. Reported-by: Aristeu Rozanski Filho <arozansk@redhat.com> Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
| * | e752x_edac: provide more info about how DIMMS/ranks are mappedMauro Carvalho Chehab2012-05-281-4/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | No funtional changes here. Only the comments got updated. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Doug Thompson <norsk5@yahoo.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
| * | i5000_edac: Fix the logic that retrieves memory informationMauro Carvalho Chehab2012-05-281-71/+79
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The logic there is broken: it basically creates two csrows for each DIMM and assumes that all DIMM's are dual rank. Only one of the csrows will contain the entire DIMM size. If single rank memories are found, they'll be marked with 0 bytes. The check if the AMB is present were also wrong. Yet, as the error reports don't use the memory size in order to credit an error to the right DIMM, that part of the driver seems to work. That's why probably nobody detected the issue yet. After this patch, the memory layout is now properly reported, when debug mode is enabled, and the number of ranks per dimm is now shown: calculate_dimm_size: ---------------------------------------------------------- calculate_dimm_size: slot 3 0 MB | 0 MB | 0 MB | 0 MB | calculate_dimm_size: slot 2 0 MB | 0 MB | 0 MB | 0 MB | calculate_dimm_size: ---------------------------------------------------------- calculate_dimm_size: slot 1 0 MB | 0 MB | 0 MB | 0 MB | calculate_dimm_size: slot 0 512 MB 1R| 512 MB 1R| 512 MB 1R| 512 MB 1R| calculate_dimm_size: ---------------------------------------------------------- calculate_dimm_size: channel 0 | channel 1 | channel 2 | channel 3 | calculate_dimm_size: branch 0 | branch 1 | (1R above means that all memories on my test machine are single-ranked) Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Cc: Doug Thompson <norsk5@yahoo.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-24 16:16:18 +0000