- 12 Jun, 2004 40 commits
-
-
Kenneth W. Chen authored
The fast system-call fall-back-path has a race: it reads PSR, modifies some bits, then writes back the new PSR. Unfortunately, the contents of PSR may change between reading and writing it. For example, an interrupt could occur which could trigger a context-switch. The context-switch might in turn flush the floating-point-high (FPH) partition to memory, clear PSR.MFH, and set PSR.DFH. To prevent this race, the patch below turns off PSR.I before reading PSR. This fixes a floating-point corruption problem that was observed on a system with a libc which has the fast system-call support enabled. The performance impact is minimal (on the order of a handful of cycles). Signed-off-by:
Kevin Tian <kevin.tian@intel.com> Signed-off-by:
Ken Chen <kenneth.w.chen@intel.com> Signed-off-by:
Asit Mallick <asit.k.mallick@intel.com> Signed-off-by:
David Mosberger <davidm@hpl.hp.com> Signed-off-by:
Linus Torvalds <torvalds@osdl.org>
-
Andrew Morton authored
I waffled over this for ages. On balance, I think it's best to mark those bh's as uptodate. And on reflection, I'm not sure why we go bringing ramdisk blockdev pages uptodate all over the place anyway. But ramdisk is weird and it passes testing. Let those dogs sleep. Signed-off-by:
Andrew Morton <akpm@osdl.org> Signed-off-by:
-
Andrew Morton authored
From: Chris Wright <chrisw@osdl.org> CHECK security/selinux/hooks.c security/selinux/hooks.c:1383:34: warning: return expression in void function security/selinux/hooks.c:3548:30: warning: return expression in void function CC security/selinux/hooks.o From: Mika Kukkonen <mika@osdl.org> Signed-off-by:
Chris Wright <chrisw@osdl.org> Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> From: Stefan Bader <shbader@de.ibm.com> tape driver changes: - Create seperate debug areas for core and discipline modules. Signed-off-by:
Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> From: Frank Pavlic <pavlic@de.ibm.com> From: Thomas Spatzier <tspat@de.ibm.com> qeth network driver changes: - Use correct request length in arp/snmp requests. - Simplify handling of empty vs. primed buffers. Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> xpram device driver changes: - Allocate request queue with blk_alloc_queue. Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> From: Cornelia Huck <cohuck@de.ibm.com> Common i/o layer changes: - Remove bogus defines. - Fix length of strncmp on bus id. Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> Cleanup s390* string functions. This replaces the 31/64 bit assembler files (strcmp[64].S, strcpy[64].S & strncpy[64].S) with a single string.c file that uses some inline assemblies to issue the string instructions. In addition some more of the generic string function got an architecture dependent implementation. Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> From: Martin Schwidefsky <schwidefsky@de.ibm.com> Introduce a TIF_SINGLE_STEP bit that causes do_debugger_trap to get called at the end of a system call. This way some code duplication in the program check handler can get removed. Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> From: Martin Schwidefsky <schwidefsky@de.ibm.com> Speedup strncpy_from_user and strnlen_from_user by using the search string instruction in the secondary space mode. Signed-off-by:
-
Andrew Morton authored
From: Martin Schwidefsky <schwidefsky@de.ibm.com> From: Guillaume Morin <guillaume@morinfr.org> This patch improves the memory detection logic. It detects any amount of holes in the memory layout up to MEMORY_CHUNK blocks of available memory. Signed-off-by:
-
Andrew Morton authored
From: Arnd Bergmann <arnd@arndb.de> This should catch almost all s390 device drivers Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Andrew Morton authored
From: Arnd Bergmann <arnd@arndb.de> Lots of trivial changes to make sparse happy on s390 arch code, mostly __user annotations. Signed-off-by:
-
Andrew Morton authored
From: Karsten Keil <kkeil@suse.de> Solves a endless IRQ loop for Tigerjet 320 based ISDN cards. Signed-off-by:
Karsten Keil <kkeil@suse.de> Signed-off-by:
-
Andrew Morton authored
From: Jorn Engel <joern@wohnheim.fh-wedel.de> Remove Volker as an NCP maintainer, he hasn't touched it in about 7 or 8 years, according to himself. Volker has acked this change. Signed-off-by:
Jörn Engel <joern@wohnheim.fh-wedel.de> Signed-off-by:
-
Andrew Morton authored
From: Hugh Dickins <hugh@veritas.com> Oleg's patch was good in that exit_mmap usually does the un-accounting; but dup_mmap still needs its own un-accounting for the case when it has charged for a vma, but error before it's inserted into child mm's list. Signed-off-by:
Hugh Dickins <hugh@veritas.com> Signed-off-by:
-
Andrew Morton authored
From: Davide Libenzi <davidel@xmailserver.org> This is a sanity check on the size parameter. Nothing explodes w/out, but the conversion to unsigned simply triggers a big allocation. Signed-off-by:
-
Andrew Morton authored
As Roland McGrath <roland@redhat.com> points out, we need to zero task->it_virt_value to prevent timer-based signal delivery, not ->it_virt_incr. Signed-off-by:
-
Andrew Morton authored
Lack of argument decls causes this warning: arch/i386/mach-generic/../mach-es7000/es7000plat.c:53: warning: function declaration isn't a prototype Let's stick the extern declaration in a header, which is where they always should be. Cc: "Brown, Len" <len.brown@intel.com> Cc: "Protasevich, Natalie" <Natalie.Protasevich@UNISYS.com> Signed-off-by:
-
Andrew Morton authored
From: Andy Whitcroft <apw@shadowen.org> Both modprobe_path and hotplug_path are arbitrarily sized at 256 bytes and that size is also expressed directly in the sysctl code. It seems reasonable to define a standard length and use that for consitancy. This patch introduces the constant KMOD_PATH_LEN and uses that. Signed-off-by:
Andy Whitcroft <apw@shadowen.org> Signed-off-by:
-
Andrew Morton authored
From: Anton Altaparmakov <aia21@cam.ac.uk> A filesystem's ->writepage() implementation nowadays must run either redirty_page_for_writepage() or the combination of set_page_writeback()/ end_page_writeback(). Failure to do so leaves the page itself marked clean but it is tagged as dirty in the radix tree (PAGECACHE_TAG_DIRTY). This incoherency can lead to all sorts of hard-to-debug problems in the filesystem like having dirty inodes at umount and losing written data. The patch updates Documentation/filesystems/Locking to reflect this requirement. Signed-off-by:
Anton Altaparmakov <aia21@cantab.net> Signed-off-by:
-
Andrew Morton authored
Reduce stack consumption in sync_inodes_sb() via read_page_state(). Signed-off-by:
-
Andrew Morton authored
Use the new read_page_state() in page-writeback.c to avoid large on-stack structures. Signed-off-by:
-
Andrew Morton authored
Use the new read_page_state() in vmscan.c to avoid large on-stack structures. Signed-off-by:
-
Andrew Morton authored
struct page_state is large (148 bytes) and we put them on the stack in awkward code paths (page reclaim...) So implement a simple read_page_state() which can be used to pluck out a single member from the all-cpus page_state accumulators. Signed-off-by:
-
Andrew Morton authored
From: Nick Piggin <nickpiggin@yahoo.com.au> Replace lots of parameters to functions in mm/vmscan.c with a structure struct scan_control. Signed-off-by:
Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by:
-
Andrew Morton authored
From: Nick Piggin <nickpiggin@yahoo.com.au> Move the invocation of ->writepage for to-be-reclaimed pages into its own function "pageout". From: Nikita Danilov <nikita@namesys.com> with small changes from Nick Piggin Signed-off-by:
Nick Piggin <nickpiggin@cyberone.com.au> Signed-off-by:
-
Andrew Morton authored
Page reclaim bales out very early if reclaim isn't working out for !__GFP_FS allocation attempts. It was a fairly arbitrary thing in the first place and chances are the caller will simply retry the allocation or will do something which is disruptive to userspace. So remove that code and do much more scanning. Signed-off-by:
-
Andrew Morton authored
Teach page reclaim to understand synchronous ->writepage implementations. If ->writepage completed I/O prior to returning we can proceed to reclaim the page without giving it another trip around the LRU. This is beneficial for ramdisk-backed S_ISREG files: we can reclaim the file's pages as fast as the ramdisk driver needs to allocate them and this prevents I/O errors due to OOM in rd_blkdev_pagecache_IO(). Signed-off-by:
-
Andrew Morton authored
NUMA-Q requires CONFIG_NUMA and CONFIG_DISCONTIGMEM to build. Cc: Martin J. Bligh <mbligh@aracnet.com> Signed-off-by:
-
Andrew Morton authored
From: William Lee Irwin III <wli@holomorphy.com> Voyager doesn't support HT, so smp_num_siblings doesn't exist. Cc: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by:
-
Andrew Morton authored
From: William Lee Irwin III <wli@holomorphy.com> Voyager doesn't compile any of the APIC or IO-APIC -related code in arch/i386/kernel/ at all -- so it's a logical impossibility that this could ever work. Cc: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by:
-
Andrew Morton authored
From: William Lee Irwin III <wli@holomorphy.com> mm/mempolicy.c: In function `verify_pages': mm/mempolicy.c:246: warning: implicit declaration of function `kmap_atomic' mm/mempolicy.c:249: warning: implicit declaration of function `kunmap_atomic' pte_offset_map() invokes kmap_atomic() via macro, without including the required header. Signed-off-by:
-
Andrew Morton authored
From: Andi Kleen <ak@suse.de> Originally by Juergen Kreileder, changed by me. The test for unhandled signals was wrong. Correct it. This helps Java users, who would see bogus printks in their kernel logs previously. Signed-off-by:
-
Andrew Morton authored
From: Arnd Bergmann <arnd@arndb.de> On s390, kmem_bufctl_t was added inside of an #ifdef, breaking 64 bit builds. Signed-off-by:
-
Andrew Morton authored
From: Andi Kleen <ak@suse.de> Fix more bad issues in the x86-64 machine check handler. - Actually test status after reading it from the register, not before. - Check the UC bit instead of the PCC bit to detect the bank which caused the exception. - Add tolerant==3 level for easier testing. This will avoid any panics. - Don't threat bank overflows as fatal - they must have come from a non fatal exception, which should not cause a panic. Signed-off-by:
-
Andrew Morton authored
From: Bruno Ducrot <poup@poupinou.org> I'm trying to replace for_each_cpu() with for_each_cpu_mask() in arch/i386/kernel/cpu/cpufreq/p4-clockmod.c Unfortunately, though, davej pointed me that for_each_cpu_mask() is not defined in -bk if CONFIG_SMP is not defined. Signed-off-by:
-
Andi Kleen authored
Discard FPU exceptions in exit and execve on x86-64 too. Signed-off-by: -
Linus Torvalds authored
sparse tokenizes everything. Including #include directives. Which means that it doesn't want to see "//" in a include filename, since that's a comment outside of a string. And inside of a string it's too dark to read.
-
Linus Torvalds authored
We need to clear all exceptions before synchronizing with the FPU, since we aren't ready to handle a FP exception here and we're getting rid of all FP state. Special thanks to Alexander Nyberg for reports and testing. Alternate patches by Sergey Vlasov and Andi Kleen, who both worked on this. Signed-off-by: Linus Torvalds <torvalds@osdl.org>
-
