- 25 Oct, 2004 40 commits
-
-
Jesse Barnes authored
On some platforms (e.g. SGI Challenge, Origin, and Altix machines), writes to I/O space aren't ordered coming from different CPUs. For the most part, this isn't a problem since drivers generally spinlock around code that does writeX calls, but if the last operation a driver does before it releases a lock is a write and some other CPU takes the lock and immediately does a write, it's possible the second CPU's write could arrive before the first's. This patch adds a mmiowb() call to deal with this sort of situation, and adds some documentation describing I/O ordering issues to deviceiobook.tmpl. The idea is to mirror the regular, cacheable memory barrier operation, wmb. Example of the problem this new macro solves: CPU A: spin_lock_irqsave(&dev_lock, flags) CPU A: ... CPU A: writel(newval, ring_ptr); CPU A: spin_unlock_irqrestore(&dev_lock, flags) ... CPU B: spin_lock_irqsave(&dev_lock, flags) CPU B: writel(newval2, ring_ptr); CPU B: ... CPU B: spin_unlock_irqrestore(&dev_lock, flags) In this case, newval2 could be written to ring_ptr before newval. Fixing it is easy though: CPU A: spin_lock_irqsave(&dev_lock, flags) CPU A: ... CPU A: writel(newval, ring_ptr); CPU A: mmiowb(); /* ensure no other writes beat us to the device */ CPU A: spin_unlock_irqrestore(&dev_lock, flags) ... CPU B: spin_lock_irqsave(&dev_lock, flags) CPU B: writel(newval2, ring_ptr); CPU B: ... CPU B: mmiowb(); CPU B: spin_unlock_irqrestore(&dev_lock, flags) Note that this doesn't address a related case where the driver may want to actually make a given write get to the device before proceeding. This should be dealt with by immediately reading a register from the card that has no side effects. According to the PCI spec, that will guarantee that all writes have arrived before being sent to the target bus. If no such register is available (in the case of card resets perhaps), reading from config space is sufficient (though it may return all ones if the card isn't responding to read cycles). I've tried to describe how mmiowb() differs from PCI posted write flushing in the patch to deviceiobook.tmpl. Signed-off-by:Jesse Barnes <jbarnes@sgi.com> Signed-off-by:
Andrew Morton <akpm@osdl.org> Signed-off-by:
Linus Torvalds <torvalds@osdl.org>
-
Paul Fulghum authored
Fix tty_io.c send_break() to assert break for proper duration. If driver break_ctl() changes task state, then break may end prematurely. USB serial driver break_ctl() sends a URB, changing task state to TASK_RUNNING. Signed-off-by:
Paul Fulghum <paulkf@microgate.com> Signed-off-by:
-
Jim Houston authored
idr_remove() should fail gracefully and warn if the id being removed is not valid. The attached patch should do the job without additional overhead. With the existing code, removing an id which was not allocated could remove a valid id which shares the same lowest layer of the radix tree. I ran a kernel with this patch but have not done any tests to force a failure. Signed-off-by:
-
Christoph Hellwig authored
Let's try to reduce the number of hand-crafted LFS checks Signed-off-by:
-
Nick Piggin authored
Add zone->all_unreclaiable to the sysrq-M and omm-killing diagnostic output. Signed-off-by:
-
Christoph Lameter authored
This is needed for an mmtimer driver update that we are currently working on. The mmtimer driver provides CLOCK_SGI_CYCLE via clock_gettime and clock_settime. With this api fix one will be able to use timer_create, timer_settime and friends from userspace to schedule and receive signals via timer interrupts of mmtimer. Changelog * Clean up timer api for drivers that use register_posix_clock. Drivers will then be able to use posix timers to schedule interrupts. * Change API for posix_clocks[].timer_create to only pass one pointer to a k_itimer structure that is now allocated and managed by the posix layer in the same way as for the other posix timer functions. * Isolate a posix_timer_event(timr) function in posix-timers.c that may be called by the interrupt routine of a timer to signal that the scheduled event has taken place. Signed-off-by:
Christoph Lameter <clameter@sgi.com> Signed-off-by:
-
Roman Zippel authored
This exports the hfs type/creator info via xattr. Signed-off-by:
Roman Zippel <zippel@linux-m68k.org> Signed-off-by:
-
Roman Zippel authored
Write back the information for a dirty resource inode directly and not via the main inode, as at the time the latter is written the former might already be gone. Signed-off-by:
-
Roman Zippel authored
Fix a small typo and read the correct time for a dir. Signed-off-by:
-
Roman Zippel authored
Manage the fs block count with a separate variable and keep a correct i_blocks for e.g. correct du output. Signed-off-by:
-
Roman Zippel authored
hfs has two dirty bits, but currently OS X uses only one of them, so match its behaviour, so that a uncleanly unmounted disk can be mounted r/w again, after it has been checked. Signed-off-by:
-
Roman Zippel authored
After a file has been renamed, the cached search key must be updated. Signed-off-by:
-
Jesse Barnes authored
Small patch to add __iomem annotations to mmtimer.c. Signed-off-by:
-
Suresh B. Siddha authored
- Remove the call to quirk_intel_irqbalance() from quirk_pciehp_msi(). - Move the x86(/x86_64)-specific quirk_intel_irqbalance() into x86 quirks.c due to its dependency on <linux/irq.h> - Mark it __init rather than __devinit, since it calls __init functions. Signed-off-by:
Suresh Siddha <suresh.b.siddha@intel.com> Signed-off-by:
-
Stephen D. Smalley authored
This patch against 2.6.9 adds a DAC ownership check to the existing MAC permission checks when setting the security.selinux attribute via setxattr. In the past, the MAC permission checks were viewed as sufficient for controlling relabeling operations, but experience in the Fedora SELinux integration has shown that a DAC check is also appropriate here, particularly under targeted policy. Signed-off-by:
Stephen Smalley <sds@epoch.ncsc.mil> Signed-off-by:
-
Randy Dunlap authored
Ignore __param section references; they aren't discarded. Error: ./drivers/mtd/devices/phram.o __param refers to 0000000000000010 R_X86_64_64 .init.text+0x0000000000000013 Error: ./drivers/scsi/dc395x.o __param refers to 0000000000000020 R_X86_64_64 .init.data+0x0000000000000064 Error: ./drivers/usb/gadget/ether.o __param refers to 0000000000000048 R_X86_64_64 .init.data+0x0000000000000020 Signed-off-by:
Randy Dunlap <rddunlap@osdl.org> Signed-off-by:
-
Jens Axboe authored
blk_recalc_rq_segments forgots to take ->max_segment_size into account and gladly merges segments bigger than we can support, thus underestimating the number of segments needed to fill it. Signed-off-by:
Jens Axboe <axboe@suse.de> Signed-off-by:
-
Rusty Russell authored
In 2.2, you used to just be able to call functions "init_module" and "cleanup_module" and they'd be magically called. These days you should use module_init(myinit)/module_exit(myexit) and avoid #ifdef MODULE. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au> Signed-off-by:
-
Rusty Russell authored
As noticed by Joey Hess (and thanks for Christoph for forwarding it). Also requirements from Werner Almesberger. If someone passes 'foo="some value"' the param engine removes the quotes and hands 'foo' and 'some value'. The __setup() parameters expect a single string, and so we try to regenerate it from the two parts. Finally, we try to place it as an environment variable. Werner wants quotes stripped out of the environment variable. It makes sense to do that for __setup, too (so it sees 'foo=some value'), since __setup functions don't usually handle quotes. Signed-off-by:
-
Rusty Russell authored
Currently, only module parameters in loaded modules are exported in /sys/modules/, while those of "modules" built into the kernel can be set by the kernel command line, but not read or set via sysfs. - move module parameters from /sys/modules/$(module_name)/$(parameter_name) to /sys/modules/$(module_name)/parameters/$(parameter_name) - remove dummy kernel_param for exporting refcnt, add "struct module *"-based attribute instead - also export module paramters for "modules" which are built into the kernel, so parameters are always accessible at /sys/modules/$(KBUILD_MODNAME)/$(parameter_name) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (modified) Signed-off-by:
Dominik Brodowski <linux@brodo.de> Signed-off-by:
-
Rusty Russell authored
There is no __attribute_unused__: use __attribute__((__unused__)). Signed-off-by:
-
Rusty Russell authored
This patch removes MODULE_PARM for everything under arch/i386. Currently only APM. Signed-off-by:
-
Rusty Russell authored
This cleans up defconfig for i386. Not much work. This patch removes MODULE_PARM for everything made by "defconfig" on x86. There are only a few left. Signed-off-by:
-
Rusty Russell authored
This patch adds a warning whenever MODULE_PARM is used. Successive patches change them over to module_param. Help appreciated! Signed-off-by:
-
Roland McGrath authored
The session leader should disassociate from its controlling terminal and send SIGHUP signals only when the whole session leader process dies. Currently, this gets done when any thread in that process dies, which is wrong. This patch fixes it. Signed-off-by:
Roland McGrath <roland@redhat.com> Signed-off-by:
-
Roland McGrath authored
This patch changes process accounting to write just one record for a process with many NPTL threads, rather than one record for each thread. No record is written until the last thread exits. The process's record shows the cumulative time of all the threads that ever lived in that process (thread group). This seems like the clearly right thing and I assume it is what anyone using process accounting really would like to see. There is a race condition between multiple threads exiting at the same time to decide which one should write the accounting record. I couldn't think of anything clever using existing bookkeeping that would get this right, so I added another counter for this. (There may be some potential to clean up existing places that figure out how many non-zombie threads are in the group, now that this count is available.) Signed-off-by:
-
Jim Hague authored
- Hardware CLUT only needs setting on pseudocolor. - Pseudo palette values need to be 32bit wide for cfb_*(). Signed-off-by:
Jim Hague <jim.hague@acm.org> Signed-off-by:
-
Jim Hague authored
- Current blanking code forces +ve (h|v)sync. Correct, match 2.4 behaviour, and introduce VESA constants for clarity. Signed-off-by:
-
Antonino Daplas authored
S3 Savage Frambuffer Driver for the following chipsets: Savage 3D Savage MX Savage 4 Savage 2000 ProSavage SuperSavage This is based from the driver written by: Denis Oliver Kropp <dok@directfb.org> Sven Neumann <neo@directfb.org> Initial Porting to 2.6 done by: Mika Pruikkonen <mpruikko@cc.hut.fi> Added the following: - Console acceleration support (imageblit, fillrect, copyarea) - Configurable - DDC2/I2C support for (ProSavage DDR-K, Savage 4 and Prosavage PM only - Configurable - 8, 16, 32 bits per pixel Signed-off-by:
Antonino Daplas <adaplas@pol.net> Signed-off-by:
-
Antonino Daplas authored
Linux framebuffer driver for Intel(R) 830M/845G/852GM/855GM/865G integrated graphics chips. Port from kernel 2.4 + some modifications and cleanup : - Fix HW accel on 845G - Use of agpgart for fb memory reservation - Add mtrr support Signed-off-by:Sylvain Meyer <sylvain.meyer@worldonline.fr> Signed-off-by:
-
Antonino Daplas authored
remove/modify all references to info->cursor Signed-off-by:
-
Antonino Daplas authored
remove/modify all references to info->cursor Signed-off-by:
-
Antonino Daplas authored
- use struct fb_pixmap pixmap instead of struct fb_pixmap sprite. The softcursor uses fb_imageblit which also uses pixmap. - remove/change all references to info->cursor Signed-off-by:
-
Antonino Daplas authored
The current cursor interface is confusing. Some fields are taken from the cursor structure in struct fb_info (enable, mask, rop fields) and the rest are taken from the passed cursor structure. These lead to a lot of confusion, making it hard for developers to write their own implementation. Also, the cursor code has several 'short-circuits', occassionally leading to undefined cursor behavior. These are the changes brought about by the patch: - Removed struct fb_cursor and related fields from struct fb_info, and instead, placed them in a struct not visible to fbdev. - The struct fb_cursor passed to fb_cursor() will _always_ contain valid data with various bitflags indicating which fields have changed - The struct fb_pixmap sprite in struct fb_info is used only by drivers with hardware cursor implementation. Initializing and allocating memory for this structure is not needed. Remove initialization and memory allocation. - The FBIO_CURSOR ioctl is broken (because fb_cursor() is broken). For now, remove fb_cursor code and make the FBIO_CURSOR ioctl always return -ENODEV. - The flag FB_CUR_SETCUR is changed to FB_CUR_SETIMAGE, indicating that the cursor sprite has changed. The image change is now checked by fbcon so drivers will not unnecessarily load the sprite image everytime. This causes hardware cursors to flicker, especially in rivafb. - Remove fb_load_cursor_image(). This is unused, and should not be implemented generically. - Documented the usage of the cursor interface in skeletonfb.c Signed-off-by:
-
Antonino Daplas authored
Alexander Kern <alex.kern@gmx.de> [PATCH] port Daniel Mantione 2.4 driver to 2.6 [PATCH] add more pci_id number [PATCH] add accelerated imgblit [PATCH] revert SDRAM_MAGIC_PLL to old behaviour [PATCH] do a "from BIOS" initialisation only by __i386__ Arnaud FONTAINE <arnaud.fontaine@free.fr> [PATCH atyfb] correction for 3D Rage Mobility L Geert Uytterhoeven <geert@linux-m68k.org> [PATCH atyfb] Atari Atyfb fixes [PATCH atyfb] Atyfb on Mach64 GX or Atari [PATCH 468] m68k sparse floating point James Simmons <jsimmons@infradead.org> [PATCH add] port to framebuffer_alloc api Nicolas Souchu <nsouch@free.fr> [PATCH] I do not found a copy, but it was incorporated too Ville Syrjälä <syrjala@sci.fi> [PATCH] fix pan with doublescan [PATCH] another double scan fix [PATCH] disable linear aperture register access [PATCH] Memory type correction [PATCH] atyfb (2.6): Fix mmio_start [PATCH] atyfb (2.6): Fix mem_refresh_rate for Mobility [PATCH] atyfb (2.6): Add RGB565 support [PATCH] atyfb: Blank LCD by turning off backlight voltage [PATCH] atyfb: Rage LT LCD register access [PATCH] atyfb: vblank irq support [PATCH] atyfb: MTRR support Antonino Daplas <adaplas@pol.net> remove/modify all references to info->cursor Signed-off-by:
-
Antonino Daplas authored
The function inter_module_get/put is to be deprecated. Remove. Signed-off-by:
-
Antonino Daplas authored
- Reduce pixmap size allocated by fbmem, i810fb and rivafb from 16-64K to 8K. This size is sufficient that a single putcs call can be accomodated by a single imageblit - Replace NR_FB_DRIVERS with FB_MAX - Trivial code, Kconfig and Documentation cleanup Signed-off-by:
-
Werner Almesberger authored
This patch updates my entry. (Long overdue ...) Signed-off-by:
Werner Almesberger <werner@almesberger.net> Signed-off-by:
-
Alasdair G. Kergon authored
Remove duplicate kfree in dm_register_target error path. Signed-Off-By:
Alasdair G Kergon <agk@redhat.com> Signed-off-by:
-
Alasdair G. Kergon authored
Remove stray semicolon. Signed-Off-By:
-
