The current read from network is working from up to down - we have some
protocol needing the data from the network, so we build the buffer space
for that protocol, add the extra space for headers and pass this buffer
down to be filled by nif get call in hope, we have guessed the incoming
packet size right. Amazingly enough this approach mostly does work, but
not always...
So, this update does work from down to up - we allocate buffer (based
on MTU or frame size info), fill it up, and pass on for upper layers.
The obvious problem is that when we should free the buffer - if at all.
In the current implementation the upper layer will free the packet on error
or when the packet is no longer needed.
While working on the issue, the additional issue did pop up - the bios
implementation does not have generic get/put interface but is using pxe
udpsend/udpreceive instead. So the udp calls are gone and undi interface
is implemented instead. Which in turn means slight other changes as we
do not need to have duplicated pxe implementation and can just use dev_net.
To align packet content, the actual read from nic is using shifted buffer by
ETHER_ALIGN (2).
Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D10232
The work to make it possible to avoid bcache via using F_NORA modifier did
miss the fact that not all loader platforms are using the bcache, and so
it is possible the modifier is not cleared, as bcache strategy function is
not used.
For fix, we make sure the checks are dont with masked flag.
This patch does fix boot for platforms which do not use bcache.
Reported by: emaste
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D10422
GNU GCC does does recognise it as a valid option and we already
use -mgeneral-regs-only that has the desired effect.
Reviewed by: emaste
Differential Revision: https://reviews.freebsd.org/D10306
While testing 32bit UEFI OVMF (which has bug about how the disk size
is presented), I did witness the errors from blkio->ReadBlocks().
It became apparent we can not entirely trust UEFI interfaces either,
so additional checks are needed.
So we use disk_ioctl(DIOCGMEDIASIZE) for disks, with fallback of
Media->LastBlock for other media.
In addition, we need to check if there is media present.
+ small fixes for error printout, and avoiding multiple blk * 512.
Reviewed by: allanjude
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D10197
As we provide the disk size verification and correction via disk_ioctl
and disk state provided by disk_open(), we can not share the partition
state in disk_devdesc structure. Also the sharing does make a lot of sense
with ufs, as only one partition is open at any given time, but zfs pools
do keep the disk devices open.
To make sure we do get the correct information about the open device,
just remove the cache.
Reviewed by: allanjude, smh
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D9757
Cleaned up some spaces where tabs should be.
efipart_ioctl() DIOCGMEDIASIZE needs to use uint64_t, not off_t
Reviewed by: allanjude
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D9562
The efipart rework did break the ARM systems as the new code is
using more exact filters to sort the devices and we need to
add support for MEDIA_FILEPATH_DP device paths.
PR: 216940
Reported by: karl@denninger.net
Reviewed by: allanjude, manu
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D9520
Fix bugs found by Coverity in efipart.c.
The Issue is that efi_devpath_last_node() can return NULL pointer, and
therefore we should check for it. In real life we really do not
expect to see it to happen, so we will just error out from the test.
CID: 1371004
Reported by: Coverity
Reviewed by: allanjude
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D9490
Rewrite EFI part device interface to present disk devices in more
user friendly way.
We keep list of three types of devices: floppy, cd and disk, the
visible names: fdX: cdX: and diskX:
Use common/disk.c and common/part.c interfaces to manage the
partitioning.
The lsdev -l will additionally list the device path.
Reviewed by: imp, allanjude
Approved by: imp (mentor), allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D8581
The arm build has revealed some of the warnings, the fix for CHAR16
warning is to switch the warning off for env.c (same as for efinet.c).
For error code we need to use macro.
Reported by: gjb
Reviewed by: imp
Approved by: imp (mentor)
Differential Revision: https://reviews.freebsd.org/D9422
Since we have dedicated libefi/env.c file for variable support, the following
changes are done:
Simple cstyle changes in env.c
Moved efi variable related commands from loader/main.c to libefi/env.c
Did create function to set "efi-version" environment variable in env.c.
This function does serve two purposes: for first a small clean up of the
loader main(), and for second, it does replace the otherwise unused
efi_variable_support hack.
A bit of cleanup of ficl backend functions. The TEST_MAIN has no meaning,
and removed few memory leaks.
The forth code is updated to use "efi-version" variable, instead of ficl
environment check.
Reviewed by: imp
Approved by: imp (mentor)
Differential Revision: https://reviews.freebsd.org/D9165
As the efi_devpath_last_node() and efi_devpath_trim() can return NULL
pointers, the consumers of this API should check the the NULL pointers.
Same for efinet_dev_init() using calloc().
Reported by: Robert Mustacchi <rm@joyent.com>
Reviewed by: jhb, allanjude
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D9203
Apparently the libstand dosfs optimization is a bit too optimistic
and did introduce possible memory corruption.
This patch is backing out the bad part and since this results in
dosfs reading full blocks now, we can also remove extra offset argument
from dv_strategy callback.
The analysis of the issue and the backout patch is provided by Mikhail Kupchik.
PR: 214423
Submitted by: Mikhail Kupchik
Reported by: Mikhail Kupchik
Reviewed by: bapt, allanjude
Approved by: allanjude (mentor)
MFC after: 1 month
Differential Revision: https://reviews.freebsd.org/D8644
lsdev command does walk over devsw list, prints list element name and
will use dv_print() callback to print the device list.
Unfortunately this approach will add unneeded noise when there are no
particular devices detected.
To remove "empty" device section headers, the dv_print() callback
should print the header instead.
In addition, fixed dv_print callback for md module.
Reviewed by: imp
Approved by: imp (mentor)
Differential Revision: https://reviews.freebsd.org/D8551
This change does modify devsw dv_print() to return the int value,
enabling walkers to interrupt the walk on non zero value from dv_print().
This will allow the pager_print actually to stop displaying data on
user input, and additionally pager is used in various *dev_print callbacks,
where it was missing.
For test, lsdev [-v] command should display data by screenfuls and should
stop when the key 'q' is pressed on pager prompt.
Reviewed by: allanjude
Approved by: allanjude (mentor)
Differential Revision: https://reviews.freebsd.org/D5461
lsdev should display detailed information about net devices only with -v
switch. This will make EFI and BIOS version of the loader to have the
same behavior.
Reviewed by: bapt, imp
Approved by: imp (mentor)
Differential Revision: https://reviews.freebsd.org/D8415
FICL definitions not in ficl/ficl32 files broke this generally. This
makes that stuff conditional on BOOT_FORTH. Also, move definitions
related to the architecture (FICL_CPUARCH and friends) into
Makefile.ficl that all parts of the tree that include files with ficl
need to include (but only if MK_FORTH == yes). In addition, had to fix
library ordering issue with LIBSTAND to keep it last. Without boot
forth, there's no references to memset to bring in memset.o from
libstand.a to satisfy libgeliboot.a's use of it. Listing libstand last
solves this issue (and it's the proper place for libstand to boot).
* On arm64 we need to use the ${MACHINE_CPUARCH} subdirectory.
* env.c is only needed when using forth so only build it there.
Sponsored by: ABT Systems Ltd
environment variable to allow conditional compilation based on EFI
being present or not. Provide efi-setenv, efi-getenv, and
efi-unsetenv, though those need improvement. Move the efi definition
to libefi (but include a reference so they get included).
files and, in a number of these places, there were problems with how they
were declared.
Some used int return instead of time_t. On some architectures the bit
width of time_t did not naturally fit into an integer and could lead to
some unexpected behavior. (For example, 32-bit ARM builds uses a 64-bit
time_t.)
Make sure the function prototypes always specify void for the argument
list when they do not have any arguemnts, otherwise some compilers can
complain about the prototype.
Reported by: Kevin Zheng
Reviewed by: sjg
Approved by: sjg (mentor)
Obtained from: Juniper Networks, Inc.
MFC after: 1 month
Differential Revision: https://reviews.freebsd.org/D7463
Add a new 'netproto' variable which can be set for now to
NET_TFTP or NET_NFS (default to NET_NONE)
From the dhcp options if one sets the root-path option to:
"ip:path", the loader will consider it is booting over NFS
(meaning same behaviour as the default current behaviour)
if the dhcp option "tftp server address" is set (option 150)
the loader will consider it is booting over tftpfs, it will then
consider the root-path options with 2 possible case
1. "path" then the IP of the tftp server will be the one passed by
the option 150, and the files will be retrieved under "path" on the tftp
server
2. "ip:path" then the IP of the tftp server will be the one passed in
the option "overwritting the IP from the option 150.
We could not "abuse" the rootpath option in the form or tftp://ip:path because
this is already used for other purpose by iPXE preventing any chainload from
iPXE to the FreeBSD loader.
Given at each open(), the loader loops over all available filesystems and keep
the "best" error, we needed to prevent tftpfs to fallback on nfs and vice versa.
the tftpfs and nfs implementation in libstand now return EINVAL early if
'netproto' for that purpose.
Reviewed by: tsoome
Relnotes: yes
Sponsored by: Gandi.net
Differential Revision: https://reviews.freebsd.org/D7628
can emulate efi_cons_poll(0 with a flag and caching the last key read with
ReadKeyStroke. This fixes the loader.efi countdown timer on Pine64 (and
other U-Boot + EFI using platforms).
Reviewed by: imp, manu
Differential Revision: https://reviews.freebsd.org/D7670
Allow netbooting on efi without having to setup any NFS server by rebuilding the
loader with LOADER_TFTP_SUPPORT like for the i386 pxeloader
Sponsored by: Gandi.net
reported by EFI implementation. This address comment on r301714.
Approved by: re (gjb), andrew (mentor)
Differential Revision: https://reviews.freebsd.org/D6787
where we assumed TERM_EMU was defined but didn't check. Fix these by also
including them under the ifdefs.
As HO is called from loader we need a null implementation so loader.efi
doesn't need to know which version of libefi it is building against.
Sponsored by: ABT Systems Ltd
code uses the GetTime function from the Runtime Service, however this has
been shown to not return a useable time on many arm64 UEFI implementations.
Reviewed by: jhb, smh
Sponsored by: ABT Systems Ltd
Differential Revision: https://reviews.freebsd.org/D6709
directly pass the request otherwise use a buffer that is a
multiple of the media size. This speeds up I/O quite a bit
when using large transfer sizes on 4Kn disks etc.
MFC after: 1 week
In r277943, the efinet_match() routine was changed to use an off by one
when matching network interfaces. The effect was that using "net1"
actually used the device attached to "net0".
Digging into the hardware that needed this workaround more, I found that
UEFI was creating two simple network protocol devices for each physical
NIC. The first device was a "raw" Ethernet device and the second device
was a "IP" device that used the IP protocol on top of the underlying
"raw" device. The PXE code in the firmware used the "IP" device to pull
across the loader.efi, so currdev was set to "net1" when booting from the
physical interface "net0". (The loaded image's device handle referenced
the "IP" device that "net1" claimed.)
However, the IP device isn't suitable for doing raw packet I/O (and the
current code to open devices exclusively actually turns the "IP" devices
off on these systems).
To fix, change the efinet driver to only attach to "raw" devices. This
is determined by fetching the DEVICE_PATH for each handle which supports
the simple network protocol and examining the last node in the path. If
the last node in the path is a MAC address, the device is assumed to be
a "raw" device and is added as a 'netX' device. If the last node is not
a MAC address, the device is ignored.
However, this causes a new problem as the device handle associated with
the loaded image no longer matches any of the handles enumerated by
efinet for systems that load the image via the "IP" device. To handle
this case, expand the logic that resolves currdev from the loaded image
in main(). First, the existing logic of looking for a handle that
matches the loaded image's handle is tried. If that fails, the device
path of the handle that loaded the loaded image is fetched via
efi_lookup_image_devpath(). This device path is then walked from the
end up to the beginning using efi_handle_lookup() to fetch the handle
associated with a path. If the handle is found and is a known handle,
then that is used as currdev. The effect for machines that load the
image via the "IP" device is that the first lookup fails (the handle
for the "IP" device isn't claimed by efinet), but walking up the
image's device path finds the handle of the raw MAC device which is used
as currdev.
With these fixes in place, the hack to subtract 1 from the unit can now
be removed, so that setting currdev to 'net0' actually uses 'net0'.
PR: 202097
Tested by: ambrisko
Sponsored by: Cisco Systems
While here, fix the various net driver callbacks to return early instead
of crashing if this fails. (The 'init' callback from the netif interface
doesn't return an error if the protocol lookup fails.)
Sponsored by: Cisco Systems
These efipart layer did several devpath related operations inline. This
just switches it over to using shared code for working with device paths.
Sponsored by: Cisco Systems
Lookup the DEVICE_PATH for each EFI network device handle and output the
string description using printf with '%S'. To honor the pager, the newline
at the end of each line is still output with pager_output().
Sponsored by: Cisco Systems
- efi_lookup_devpath() uses the DEVICE_PATH_PROTOCOL to obtain the
DEVICE_PATH for a given EFI handle.
- efi_lookup_image_devpath() uses the LOADED_IMAGE_DEVICE_PATH_PROTOCOL
to lookup the device path of the device used to load a loaded image.
- efi_devpath_name() uses the DEVICE_PATH_TO_TEXT_PROTOCOL to generate
a string description of a device path. The returned string is a CHAR16
string that can be printed via the recently added '%S' format in
libstand's printf(). Note that the returned string is returned in
allocated storage that should be freed by calling
efi_free_devpath_name().
- efi_devpath_last_node() walks a DEVICE_PATH returning a pointer to the
final node in the path (not counting the terminating node). That is,
it returns a pointer to the last meaninful node in a DEVICE_PATH.
- efi_devpath_trim() generates a new DEVICE_PATH from an existing
DEVICE_PATH. The new DEVICE_PATH does not include the last
non-terminating node in the original path. If the original DEVICE_PATH
only contains the terminating node, this function returns NULL.
The caller is responsible for freeing the returned DEVICE_PATH via
free().
- efi_devpath_handle() attempts to find a handle that corresponds to a
given device path. However, if nodes at the end of the device path do
not have valid handles associated with them, this function will return
a handle that matches a node earlier in the device path. In particular,
this function returns a handle for the node closest to the end of the
device path which has a valid handle.
Sponsored by: Cisco Systems
return value when it could return 1 (indicating we should stop).
Fix a few instances of pager_open() / pager_close() not being called.
Actually use these routines for the environment variable printing code
I just committed.
Coverity reports an uninitialized "len" in case the switch defaults
without hitting any case. Respect the original intent and quell the
false positive with the relatively new __unreachable() builtin.
CID: 1347796
Don't crash if the user has more than 31 of them. A follow-up to
r298230.
Reviewed by: allanjude
Relnotes: maybe
Sponsored by: EMC / Isilon Storage Division
Differential Revision: https://reviews.freebsd.org/D6212
The new bcache code does not know the size of the disk, and therefore may attempt to read past the end of the disk while trying to fill its read-ahead cache.
This is usually not an issue, it fails gracefully on all of my machines, but some BIOSes seem to retry the reads for up to 30 seconds each, resulting in a long stall during boot
Submitted by: Toomas Soome <tsoome@me.com>
Reviewed by: jhb, np
Differential Revision: https://reviews.freebsd.org/D6109
The block cache implementation in loader has proven to be almost useless, and in worst case even slowing down the disk reads due to insufficient cache size and extra memory copy.
Also the current cache implementation does not cache reads from CDs, or work with zfs built on top of multiple disks.
Instead of an LRU, this code uses a simple hash (O(1) read from cache), and instead of a single global cache, a separate cache per block device.
The cache also implements limited read-ahead to increase performance.
To simplify read ahead management, the read ahead will not wrap over bcache end, so in worst case, single block physical read will be performed to fill the last block in bcache.
Booting from a virtual CD over IPMI:
0ms latency, before: 27 second, after: 7 seconds
60ms latency, before: over 12 minutes, after: under 5 minutes.
Submitted by: Toomas Soome <tsoome@me.com>
Reviewed by: delphij (previous version), emaste (previous version)
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D4713
