This patch adds full cell iommu support (and iommu disabled mode).
It implements mapping/unmapping of iommu pages on demand using the
standard powerpc iommu framework. It also supports running with
iommu disabled for machines with less than 2GB of memory. (The
default is off in that case, though it can be forced on with the
kernel command line option iommu=force).
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Now that the direct DMA ops supports an offset, we use that instead
of defining our own.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch makes dma_alloc_coherent() use node local allocation when
using the direct DMA ops. The node is obtained from the new device
extension. If no such extension is present, the current node is used.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch adds an optional global offset that can be added to DMA addresses
when using the direct DMA operations.
That brings it a step closer to the 32 bits direct DMA operations, and makes
it useable on Cell when the MMU is disabled and we are using a spider
southbridge.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
We use the powerpc specific low level MMIO accessor variants instead
of readl() or readl_be() because we know spidernet is not a real PCI
device and we can thus avoid the performance hit caused by the PCI
workarounds.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch implements a workaround for a Spider PCI host bridge bug
where it doesn't enforce some of the PCI ordering rules unless some
manual manipulation of a special register is done. In order to be
fully compliant with the PCI spec, I do this on every MMIO read
operation.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch reworks the way iSeries hooks on PCI IO operations (both MMIO
and PIO) and provides a generic way for other platforms to do so (we
have need to do that for various other platforms).
While reworking the IO ops, I ended up doing some spring cleaning in
io.h and eeh.h which I might want to split into 2 or 3 patches (among
others, eeh.h had a lot of useless stuff in it).
A side effect is that EEH for PIO should work now (it used to pass IO
ports down to the eeh address check functions which is bogus).
Also, new are MMIO "repeat" ops, which other archs like ARM already had,
and that we have too now: readsb, readsw, readsl, writesb, writesw,
writesl.
In the long run, I might also make EEH use the hooks instead
of wrapping at the toplevel, which would make things even cleaner and
relegate EEH completely in platforms/iseries, but we have to measure the
performance impact there (though it's really only on MMIO reads)
Since I also need to hook on ioremap, I shuffled the functions a bit
there. I introduced ioremap_flags() to use by drivers who want to pass
explicit flags to ioremap (and it can be hooked). The old __ioremap() is
still there as a low level and cannot be hooked, thus drivers who use it
should migrate unless they know they want the low level version.
The patch "arch provides generic iomap missing accessors" (should be
number 4 in this series) is a pre-requisite to provide full iomap
API support with this patch.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch makes the Cell DMA code work on both the Spider and the Axon
south bridges by turning cell_dma_valid into a variable instead of a
constant. This is a temporary patch until we have full iommu support.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
When enabled in Kconfig, it will pick up any of_platform_device
matching it's match list (currently type "pci", "pcix", "pcie",
or "ht" and setup a PHB for it.
Platform must provide a ppc_md.pci_setup_phb() for it to work
(for doing the necessary initialisations specific to a given PHB
like setting up the config space ops).
It's currently only available on 64 bits as the 32 bits PCI code
can't quite cope with it in it's current form. I will fix that
later.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Add a "parent" struct device to our PCI host bridge data structure so that
PCI can be rooted off another device in sysfs.
Note that arch/ppc doesn't use it, only arch/powerpc, though it's available
for both 32 and 64 bits.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The BUID is the first entry of a PCI host bridge "reg" property.
Now that PCI busses can be anywhere in the device-tree, we need to
fully translate the value there to a CPU physical address before
we can use it with RTAS.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
When parsing the OF "ranges" properties of PCI host busses to determine
the mapping of a PCI bus, we need to translate the "parent" address using
the prom_parse.c routines in order to obtain a CPU physical address.
This wasn't necessary while PCI busses were always at the root of the
device-tree but this is no longer the case on Cell where they can be
anywhere in the tree.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch adds a bus device notifier to the of_platform bus type on
cell to setup the DMA operations for of_platform_devices. We currently
use the PCI operations as Cell use a special version of them that
happens to be suitable for our needs.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch completely refactors DMA operations for 64 bits powerpc. 32 bits
is untouched for now.
We use the new dev_archdata structure to add the dma operations pointer
and associated data to struct device. While at it, we also add the OF node
pointer and numa node. In the future, we might want to look into merging
that with pci_dn as well.
The old vio, pci-iommu and pci-direct DMA ops are gone. They are now replaced
by a set of generic iommu and direct DMA ops (non PCI specific) that can be
used by bus types. The toplevel implementation is now inline.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Remove the now unused "liobn" field in ibmveth which also avoids
having insider knowledge of the iommu table in that driver.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Santiago Leon <santil@us.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Hook up of_platform_bus_probe with the cell platform in order to publish
the non-PCI devices in the device-tree of cell blades as of_platform_device(s)
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch first splits of_device.c and of_platform.c, the later containing
the bits relative to of_platform_device's. On the "breaks" side of things,
drivers uisng of_platform_device(s) need to include asm/of_platform.h now
and of_(un)register_driver is now of_(un)register_platform_driver.
In addition to a few utility functions to locate of_platform_device(s),
the main new addition is of_platform_bus_probe() which allows the platform
code to trigger an automatic creation of of_platform_devices for a whole
tree of devices.
The function acts based on the type of the various "parent" devices encountered
from a provided root, using either a default known list of bus types that can be
"probed" or a passed-in list. It will only register devices on busses matching
that list, which mean that typically, it will not register PCI devices, as
expected (since they will be picked up by the PCI layer).
This will be used by Cell platforms using 4xx-type IOs in the Axon bridge
and can be used by any embedded-type device as well.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Add support for southbridges using the MPIC interrupt controller to
the native cell platforms.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch applies on top of the MPIC DCR support. It makes the MPIC
driver capable of a lot more auto-configuration based on the device-tree,
for example, it can retreive it's own physical address if not passed as
an argument, find out if it's DCR or MMIO mapped, and set the BIG_ENDIAN
flag automatically in the presence of a "big-endian" property in the
device-tree node.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch implements support for DCR based MPIC implementations. Such
implementations have the MPIC_USES_DCR flag set and don't use the phys_addr
argument of mpic_alloc (they require a valid dcr mapping in the device node)
This version of the patch can use a little bif of cleanup still (I can
probably consolidate rb->dbase/doff, at least once I'm sure on how the
hardware is actually supposed to work vs. possible simulator issues) and
it should be possible to build a DCR-only version of the driver. I need
to cleanup a bit the CONFIG_* handling for that and probably introduce
CONFIG_MPIC_MMIO and CONFIG_MPIC_DCR.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch makes the EMAC driver use the new DCR access methods. It
doesn't yet uses dcr_map() and thus still only work with real DCRs.
This will be fixed in a later patch
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch adds new dcr_map/dcr_read/dcr_write accessors for DCRs that
can be used by drivers to transparently address either native DCRs or
memory mapped DCRs. The implementation for memory mapped DCRs is done
after the binding being currently worked on for SLOF and the Axon
chipset. This patch enables it for the cell native platform
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
These were inherited from ARCH=ppc, but are not needed since parsing of interrupts
should be done via the of_* functions (who can do swizzling). If we ever need to
do non-standard swizzling on bridges without a device-node, then we might add
back a slightly different version of ppc_md.pci_swizzle but for now, that is not
the case.
I removed the couple of calls for these in 83xx. If that breaks something, then
there is a problem with the device-tree on these.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch reworks the way IRQs are fixed up on PCI for arch powerpc.
It makes pci_read_irq_line() called by default in the PCI code for
devices that are probed, and add an optional per-device fixup in
ppc_md for platforms that really need to correct what they obtain
from pci_read_irq_line().
It also removes ppc_md.irq_bus_setup which was only used by pSeries
and should not be needed anymore.
I've also removed the pSeries s7a workaround as it can't work with
the current interrupt code anyway. I'm trying to get one of these
machines working so I can test a proper fix for that problem.
I also haven't updated the old-style fixup code from 85xx_cds.c
because it's actually buggy :) It assigns pci_dev->irq hard coded
numbers which is no good with the new IRQ mapping code. It should
at least use irq_create_mapping(NULL, hard_coded_number); and possibly
also set_irq_type() to set them as level low.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Allow architectures to provide their own implementation of the big endian MMIO
accessors and "repeat" MMIO accessors for use by the generic iomap.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
More-or-less-tested-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The following patch fixes the failure of sunhme drivers on x86 hosts
due to missing pci_enable_device() and pci_set_master() calls, lost
during code refactoring. It has been filed as bugzilla bug #7502 [0]
and Debian bug #397460 [1].
[0] http://bugzilla.kernel.org/show_bug.cgi?id=7502
[1] http://bugs.debian.org/397460
Signed-off-by: Jurij Smakov <jurij@wooyd.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's bitrotten, long unmaintained, long hidden under BROKEN_ON_SMP,
etc. As scheduled in feature-removal-schedule.txt, and ack'd several
times on lkml.
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Since we never checked the ->family value of templates
before, many applications simply leave it at zero.
Detect this and fix it up to be the pol->family value.
Also, do not clobber xp->family while reading in templates,
that is not necessary.
Signed-off-by: David S. Miller <davem@davemloft.net>
With async scanning, we're now tripping the BUG_ON in
sas_ex_discover_end_dev(), so make the error handling here correct.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This replaces the linear search algorithm for reverse lookup with
binary search.
It has the advantage of better scalability: O(log2(N)) instead of O(N).
This means that the average number of iterations is reduced from 250
(linear search if each value appears equally likely) down to at most 9.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This patch deprecates the existing use of an arbitrary value TFRC_SMALLEST_P
for low-threshold values of p. This avoids masking low-resolution errors.
Instead, the code now checks against real boundaries (implemented by preceding
patch) and provides warnings whenever a real value falls below the threshold.
If such messages are observed, it is a better solution to take this as an
indication that the lookup table needs to be re-engineered.
Changelog:
----------
This patch
* makes handling all TFRC resolution errors local to the TFRC library
* removes unnecessary test whether X_calc is 'infinity' due to p==0 -- this
condition is already caught by tfrc_calc_x()
* removes setting ccid3hctx_p = TFRC_SMALLEST_P in ccid3_hc_tx_packet_recv
since this is now done by the TFRC library
* updates BUG_ON test in ccid3_hc_tx_no_feedback_timer to take into account
that p now is either 0 (and then X_calc is irrelevant), or it is > 0; since
the handling of TFRC_SMALLEST_P is now taken care of in the tfrc library
Justification:
--------------
The TFRC code uses a lookup table which has a bounded resolution.
The lowest possible value of the loss event rate `p' which can be
resolved is currently 0.0001. Substituting this lower threshold for
p when p is less than 0.0001 results in a huge, exponentially-growing
error. The error can be computed by the following formula:
(f(0.0001) - f(p))/f(p) * 100 for p < 0.0001
Currently the solution is to use an (arbitrary) value
TFRC_SMALLEST_P = 40 * 1E-6 = 0.00004
and to consider all values below this value as `virtually zero'. Due to
the exponentially growing resolution error, this is not a good idea, since
it hides the fact that the table can not resolve practically occurring cases.
Already at p == TFRC_SMALLEST_P, the error is as high as 58.19%!
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This
* adds documentation about the lowest resolution that is possible within
the bounds of the current lookup table
* defines a constant TFRC_SMALLEST_P which defines this resolution
* issues a warning if a given value of p is below resolution
* combines two previously adjacent if-blocks of nearly identical
structure into one
This patch does not change the algorithm as such.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
1) For the forward X_calc lookup, it
* protects effectively against RTT=0 (this case is possible), by
returning the maximal lookup value instead of just setting it to 1
* reformulates the array-bounds exceeded condition: this only happens
if p is greater than 1E6 (due to the scaling)
* the case of negative indices can now with certainty be excluded,
since documentation shows that the formulas are within bounds
* additional protection against p = 0 (would give divide-by-zero)
2) For the reverse lookup, it warns against
* protects against exceeding array bounds
* now returns 0 if f(p) = 0, due to function definition
* warns about minimal resolution error and returns the smallest table
value instead of p=0 [this would mask congestion conditions]
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This fixes the following small error in tfrc_calc_x_reverse_lookup.
1) The table is generated by the following equations:
lookup[index][0] = g((index+1) * 1000000/TFRC_CALC_X_ARRSIZE);
lookup[index][1] = g((index+1) * TFRC_CALC_X_SPLIT/TFRC_CALC_X_ARRSIZE);
where g(q) is 1E6 * f(q/1E6)
2) The reverse lookup assigns an entry in lookup[index][small]
3) This index needs to match the above, i.e.
* if small=0 then
p = (index + 1) * 1000000/TFRC_CALC_X_ARRSIZE
* if small=1 then
p = (index+1) * TFRC_CALC_X_SPLIT/TFRC_CALC_X_ARRSIZE
These are exactly the changes that the patch makes; previously the code did
not conform to the way the lookup table was generated (this difference resulted
in a mean error of about 1.12%).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This adds documentation for the TCP Reno throughput equation which is at
the heart of the TFRC sending rate / loss rate calculations.
It spells out precisely how the values were determined and what they mean.
The equations were derived through reverse engineering and found to be
fully accurate (verified using test programs).
This patch does not change any code.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This avoids a (harmless) warning message being printed at the DCCP server
(the receiver of a DCCP half connection).
Incoming packets are both directed to
* ccid_hc_rx_packet_recv() for the server half
* ccid_hc_tx_packet_recv() for the client half
The message gets printed since on a server the client half is currently not
sending data packets.
This is resolved for the moment by checking the DCCP-role first. In future
times (bidirectional DCCP connections), this test may have to be more
sophisticated.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
The main object of this patch is the following bug:
==> In ccid3_hc_tx_packet_recv, the parameters p and X_recv were updated
_after_ the send rate was calculated. This is clearly an error and is
resolved by re-ordering statements.
In addition,
* r_sample is converted from u32 to long to check whether the time difference
was negative (it would otherwise be converted to a large u32 value)
* protection against RTT=0 (this is possible) is provided in a further patch
* t_elapsed is also converted to long, to match the type of r_sample
* adds a a more debugging information regarding current send rates
* various trivial comment/documentation updates
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This bug resulted in ccid3_hc_tx_send_packet returning negative
delay values, which in turn triggered silently dequeueing packets in
dccp_write_xmit. As a result, only a few out of the submitted packets made
it at all onto the network. Occasionally, when dccp_wait_for_ccid was
involved, this also triggered a bug warning since ccid3_hc_tx_send_packet
returned a negative value (which in reality was a negative delay value).
The cause for this bug lies in the comparison
if (delay >= hctx->ccid3hctx_delta)
return delay / 1000L;
The type of `delay' is `long', that of ccid3hctx_delta is `u32'. When comparing
negative long values against u32 values, the test returned `true' whenever delay
was smaller than 0 (meaning the packet was overdue to send).
The fix is by casting, subtracting, and then testing the difference with
regard to 0.
This has been tested and shown to work.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
The TFRC nofeedback timer normally expires after the maximum of 4
RTTs and twice the current send interval (RFC 3448, 4.3). On LANs
with a small RTT this can mean a high processing load and reduced
performance, since then the nofeedback timer is triggered very
frequently.
This patch provides a configuration option to set the bound for the
nofeedback timer, using as default 100 milliseconds.
By setting the configuration option to 0, strict RFC 3448 behaviour
can be enforced for the nofeedback timer.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
