Move autogen.sh invocation of libjaylink from the configure to the
bootstrap script because the configure script is included in tarball
releases but autotools are not required to be available on end-user
machines.
Thanks to Paul Fertser for spotting this.
Change-Id: I5489ae83885157a01803eed51a7328e47d67ea6d
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3569
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
The J-Link driver is not JTAG only and the correct spelling of the
company is "SEGGER".
Change-Id: Ic1062247d3bf0d564960f1ac6dfc86ffb34593e0
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3530
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Include libjaylink in DIST_SUBDIRS only if J-Link driver is enabled.
Change-Id: Ib32bb6a5d8fe9bb6e93d968802733eb4973f2c25
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3529
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Do not configure internal libjaylink if libusb-1.0 is not available or
if J-Link driver is disabled (--disable-jlink).
Change-Id: I021bca91dbbc33888a997c664f7836225306c3ef
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3528
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Since OpenOCD doesn't fit most common definitions of the word "daemon",
using it in the documentation is confusing.
Reported by IRC user ohsix.
Change-Id: I688d722771b084b17c2a7af8e83fd64bab6141b8
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/3634
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Since merge of #3149 OpenOCD start with an unresponsive SAMD or SAM4L
resulted in segfaults. First was in cortex_m_assert_reset
(fixed by #3552), second was in samd_handle_reset_deassert()
/sam4l_handle_reset_deassert().
The change replaces mem_ap_write_u32/8 by target_write_u32/8.
It also takes better care about examining and polling target before
debug control registers are set. It prevents lockup when 'reset halt'
is issued on unresponsive cpu.
Change-Id: I2516489f4771aebfc1118d174f527497b8a201ad
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/3603
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
The MDR parts have two kinds of memory: main (regular memory-mapped
memory for code and data) and "info" (not memory-mapped). When OpenOCD
is requested to erase the info memory block, it should do just that,
instead of erasing everything including main memory.
Change-Id: I498142ca50d4a7b669b7776180b0dbcea63a5328
Reported-by: Eldar Khayrullin <eldar.khayrullin@mail.ru>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/3853
Tested-by: jenkins
Reviewed-by: Eldar Khayrullin <eldar.khayrullin@mail.ru>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
On multi-core systems, with some cores in power-down state, examination
will fail for these cores. Make sure assert- and deassert_reset functions
don't crash due to uninitialized variables.
Change-Id: I472f8d19af2cd3c770c05f3e57a31b35a863b687
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/3552
Tested-by: jenkins
Reviewed-by: Jiri Kastner <cz172638@gmail.com>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
At least on my (phyCORE-AM335X) system, the AM335x watchdog
needs to be disabled to use OpenOCD for more than 6.5 seconds
after reset.
Change-Id: I3d883a9f572b0ccb92f9864853a00c372e39d7f2
Signed-off-by: Harald Welte <laforge@gnumonks.org>
Reviewed-on: http://openocd.zylin.com/3391
Tested-by: jenkins
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Add semihosting support for ARMv7-A based processors.
Tested with custom Vybrid VF610 based board
and Pandaboard ES (Rev. B1) board (Cortex-A9).
Change-Id: I6b896a61c1c6a1c5dcf89de834486f82dd6c80a2
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Tsung-Han Lin <tsunghan.tw@gmail.com>
Reviewed-on: http://openocd.zylin.com/2908
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Nand write command :
nand_fileio_cleanup() always returns ERROR_OK. Due to this,
handle_nand_write_command() retuns ERROR_OK in the case
of nand failure. ERROR_FAIL should be returned.
Flash erase_sector command :
handle_flash_erase_command() always returns ERROR_OK even if
the erase functionality of actual driver implementation fails.
retval value should be returned.
Flash write_bank command :
handle_flash_write_bank_command() returns ERROR_OK even if
fileio_open() and fileio_read fails. ERROR_FAIL should be
returned.
Load_image command :
handle_load_image_command() retuns ERROR_OK even if image_open()
fails. ERROR_FAIL should be returned.
When the buffer is null, breaking the loop without setting
retval = ERROR_FAIL would cause load_image to return ERROR_OK.
Change-Id: Ice32f6036971ab5e8e4dd65edf54b394b001c80c
Signed-off-by: HarishKumar <harishpresent@gmail.com>
Reviewed-on: http://openocd.zylin.com/2431
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Instead of updating these regularly we can just accept any IDCODE for the
boundary scan TAP.
The only downside might be that it's not immediately obvious if you
source a config for the wrong type of STM32.
Change-Id: I96d4d81699a491b3a46de3f0d3fd078ffddad4e4
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/3385
Tested-by: jenkins
Both boards use SEGGER interface.
nRF52 flash driver is in change 3511.
The board files were needed for transport select to SWD. The default
jtag transport didn't failover to SWD.
Change-Id: Id9c72aa88eeeb6a4e1999ee182f5284dbe535c05
Signed-off-by: Rick Foos <rfoos@solengtech.com>
Reviewed-on: http://openocd.zylin.com/3600
Tested-by: jenkins
Reviewed-by: Fredrik Hederstierna <fredrik.hederstierna@gmail.com>
Reviewed-by: Michael Dietz <mjdietzx@gmail.com>
Reviewed-by: Jiří Pinkava <j-pi@seznam.cz>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
With -Og gcc doesn't perform as many optimizations, and as a result
warns about some code that it wouldn't otherwise warn about.
These fixes all assign values to otherwise uninitialized variables.
Change-Id: I9a6ea5eadd73673891ecfec568b8b00d78b596a5
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: http://openocd.zylin.com/3779
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
To speed up downloads, OpenOCD sends gdb OK when a write is received,
even before the write has actually occurred. The failure is then
returned for the next write. That leads to the following confusing
behavior:
```
(gdb) p/x *((int*)0xdeadbeef)=8675309
$2 = 0x845fed
(gdb) p/x *((int*)0x80000000)=6874742
Cannot access memory at address 0x80000000
```
While it's actually the first write that failed.
This change hacks around this problem by not sending OK for small writes
(len<8) until the write has actually occurred. This does not impact
download speed, since during downloads (almost) all writes will have
much larger length.
Change-Id: I1f8b9bb19b0707487f840df6871e372e4ba228dd
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: http://openocd.zylin.com/3803
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Known big flash parts such as LPC11u68/e68 have a non-uniform memory
organisation, the first 24 sectors are 4k, the rest are 32k.
Change-Id: Icf515152dfc54ec0ca187561d2d63088b9640f14
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/3802
Tested-by: jenkins
Reviewed-by: akaWolf
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
struct flash_driver has a default_padded_value field that is similar,
but it can be changed by the user for the specific purpose of padding.
Add a new erased_value field and initialize it for all targets,
particularly stm32lx, xmc4xxx and virtual.
Use this value in core.c:default_flash_mem_blank_check(), the slow path.
Extend the target API to pass erased_value down to target code.
Adding an argument ensures that we catch all callers.
This allows us to merge xmc4xxx.c:xmc4xxx_blank_check_memory() into
armv7m:armv7m_blank_check_memory().
It further allows us to use default_flash_blank_check() in place of
xmc4xxx.c:xmc4xxx_flash_blank_check(), adding a potential slow path
fallback, as well as stm32lx:stm32lx_erase_check(), adding the potential
armv7m fast path with fallback to default_flash_mem_blank_check().
Fix a mips32 code comment while at it (zeroed -> erased).
The armv4_5 and mips32 target implementations will now error out if an
erase value other than 0xff is used, causing default_flash_blank_check()
to fall back to the default_flank_mem_blank_check() slow path.
Change-Id: I39323fbbc4b71c256cd567e439896d0245d4745f
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3497
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
The current implementation is not suitable for user provided data
because it does not detect invalid inputs in many cases. For example,
the string "aa0xbb" is successfully converted to the 3 bytes: 0xaa,
0x00 and 0xbb. An other example is "aabi" which is successfully
converted to the 2 bytes: 0xaa and 0x0b. Both are obviously incorrect.
Make unhexify() robust on invalid data and use more appropriate data
types for its parameters. Also, add a small documentation for the
function.
Change-Id: Idb799beb86fc608b066c8a76365021ed44c7f890
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3792
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Clean up some type casts and misuses of format specifiers in preparation
for target address type changes.
Change-Id: Idf08286f41bca636e35a09e8ddc1d71af3d6e151
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3717
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Insert a space before parenthesis in logs that we will need to touch
for 64-bit target addresses.
While at it, do a couple more surrounding whitespace fixes.
Change-Id: I1080c0470aab51cf7bd56e67e934344d0bf4c5c1
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3716
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
A space after the format specifier was missing.
Change-Id: Ib67eb0fb0d6e05d765206d30d5e4a74cb41bb47b
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3715
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
working_area_phys_spec clearly refers to the physical, not virtual address.
Change-Id: I639ea00bb5d05e845b8a56815a571375849f1225
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3714
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
This makes it possible to run OpenOCD as unprivileged user on FreeBSD.
Change-Id: If1182d09ad2d51f370ae7e6da02f19c7a21c66c9
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/3800
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Even when TRST and SRST are not present in reset_config we still should
set them appropriately (to 1) as we can't tristate them anyhow.
Change-Id: Iec5bcf09340136f5e6ccfb05fa2697c53fa6609f
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/3798
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Commit 6d5b4d709c causes failed assert
when JTAG transport is selected:
src/target/arm_adi_v5.h:346: dap_queue_ap_read:
Assertion `ap->dap->ops != ((void *)0)' failed.
As check_security runs early in examine-start event, dap->ops must
be checked.
Change-Id: Ibd8312a3c668fbce834eed9790eabeed794117aa
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/3712
Tested-by: jenkins
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The Linksys WAG200G router has a TI AR7 SoC and a 4MB Flash layout
similar to the Netgear DG834v3 router [1].
Below is an example of a successful flash recovery, using a TUMPA,
connected to the MIPS EJTAG 2.6 header (JP102) of the router.
WAG200G [2] TUMPA [3]
Desc Pin Pin Desc
------------------------
nTRST 1 3 nTRST
TDI 3 5 TDI
TDO 5 13 TDO
TMS 7 7 TMS
TCK 9 9 TCK
nSRST 11 15 RST
GND 2 4 GND
Note that nSRST is optional to halt the CPU, but is required to probe
the flash. For instance, recover the kernel with:
$ sudo ./src/openocd -s tcl \
-f interface/ftdi/tumpa.cfg \
-f tools/firmware-recovery.tcl \
-c 'board linksys-wag200g;
reset_config srst_only;
flash_part kernel /path/to/kernel.bin;
shutdown'
[1] https://wiki.openwrt.org/toh/linksys/wag200g
[2] https://www.linux-mips.org/wiki/JTAG#JTAG_headers
[3] http://www.tiaowiki.com/w/TIAO_USB_Multi_Protocol_Adapter_User's_Manual#20_PIN_JTAG_Connector
Change-Id: I952ba9f706e2e4f8f95ca03a5fa58f391ca030b6
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-on: http://openocd.zylin.com/3776
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
As per the documentation, used "disabled" as the value to disable, as this
is the same value to disable the telnet and tcl server.
Change-Id: Idc4a8580098ec1107dcc6e1f59e817ecdebc38ac
Signed-off-by: Sandeep Mistry <s.mistry@arduino.cc>
Reviewed-on: http://openocd.zylin.com/3175
Tested-by: jenkins
Reviewed-by: Cristian Maglie
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Update to latest libjaylink version and incorporate API changes.
The major change in libjaylink is a rework of the device discovery.
Please test device discovery extensively, especially with multiple
devices connected. All other changes in libjaylink are of minor
importance for OpenOCD.
Change-Id: I9a50e83f59557505ce29809c7762c5df1cec10eb
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3735
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Steven Stallion <stallion@squareup.com>
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Currently, the log verbosity of libjaylink is not accessible from the
user interface. Therefore, changing the log verbosity of libjaylink is
only possible from within the J-Link driver source code which is not
acceptable for end users.
Output the libjaylink log messages through the logging module of
OpenOCD rather than directly to stderr.
Change-Id: I6bf7bf8f4c8a12fb9e955eeced68224545fa0b5c
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3701
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Add log_vprintf_lf() to enable the possibility to output log messages
with a variable argument list.
Change-Id: I7fd6e93db63a7d98f662df2881a42e4d923c3848
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3709
Tested-by: jenkins
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Cortex-M7 has autoincrement range only 1024 bytes,
surprisingly smaller than M3, M4.
Change-Id: I35ff1f0e093aac4af79f98eb3b8058d4295942d1
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/3737
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Large flash parts time out when unlocking. Mass erase time is specified
as 32 seconds worst case for some parts. Automatic mass erase is
probably not using x8 parallelism, though, but a too large timeout
shouldn't hurt.
Also, use the new define instead of hardcoded timeout when mass
erasing.
Change-Id: Ib5af60d52ed7d53277bfe7176c4c44f79d3a26bc
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/3738
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
* port to new migen
* streamline package/part specification
* add pullup (Series3, Series6) and pullnone (Series7) for unused pins
as xilinx impact/vivado do it.
* specify respective toolchains
* build Series7 with vivado (broader support, faster)
* point to prebuilt bitstreams at https://github.com/jordens/bscan_spi_bitstreams
Change-Id: Ibfef3d78f855b754425f3e6131e2e49fa111e09a
Signed-off-by: Robert Jordens <jordens@gmail.com>
Reviewed-on: http://openocd.zylin.com/3173
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Reviewed-by: Robert Jördens
Reviewed-by: William D. Jones
Reviewed-by: Tim "mithro" Ansell <mithro@mithis.com>
The code memory size was a bug and seemed to be working by
accident since there happened to be 256 pages in the device that was
tested on which corresponded to 256kb.
Also don't fail if memory size != expected memory size based on hwid
as this hwid is unstable and should be used only for debug/diagnostics.
Change-Id: I4e98f7498a36c53fc51783eddfdaba704d30e3ca
Signed-off-by: Michael Dietz <mjdietzx@gmail.com>
Reviewed-on: http://openocd.zylin.com/3510
Tested-by: jenkins
Reviewed-by: Jiří Pinkava <j-pi@seznam.cz>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Currently if ftdi device is removed, OpenOCD will stall forever.
Only kill -9 will help in this case.
This patch makes use of libusb timeout functions and
trying to break out of while loop if some error is detected.
[andreas.fritiofson@gmail.com]: Add missing retval check
Change-Id: I97506190e376026705f14ef9fe37dc811b99b3ac
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/3419
Reviewed-by: Andreas Färber <afaerber@suse.de>
Tested-by: jenkins
Believe in using defines to make maintenance easier.
Change-Id: I8edf151352131bbf2b884dfcd67ca5764b11b13c
Signed-off-by: Kent Brinkley <jkbrinkley.imgtec@gmail.com>
Reviewed-on: http://openocd.zylin.com/2350
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Added new chip id based flash size.
Change-Id: I5b5e71074af0e50352443f66f88adfc6e14280bf
Signed-off-by: Peter Kuhar <peter@pkuhar.com>
Reviewed-on: http://openocd.zylin.com/3732
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
arm-none-eabi target triplet defaults to Little Endian, and so far any
submitted machine code snippets have been verified to be Little Endian.
However a user might override [ARM_]CROSS_COMPILE with an armeb toolchain,
potentially resulting in invalid machine code.
Let's be safe and enforce Little Endian mode for assembler and compiler.
Change-Id: I9cefe24689eaded25d60ffb1f254b254e8d76f9d
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3498
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Kinetis family employs strange concept of Flash Configuration Field at
address 0x400 of program flash. Writing incorrect data to FCF may
permanently lock the device.
The change introduces 'kinetis fcf_source protection' mode. In this mode
write of flash image data to FCF is prevented. FCF data build from
protection (set by 'flash protect' command) are written instead.
FCF data are written also just after erase of relevant sector. It
protects device from locking security by reset or power cycle after erase.
prot_blocks array is used as protection blocks have bigger size than sectors.
Alignment and padding programming sections is rewritten to fix
writing with not section boundary aligned begin.
Change-Id: I9fc8bd37d6f627fb8ed7abb7f7560e78a740b195
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/3562
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Add kinetis_ftfx_decode_error() to show flash error type in human
readable message.
Add kinetis_ftfx_prepare() to prepare flash module just once in
command (not each time kinetis_ftfx_command() is called).
Change target_read/write_memory() to target_read/write_u8/32().
Make ftfx_fstat parameter of kinetis_ftfx_command() optional.
Longword flash write:
Fix huge memory leak after write of unaligned block.
Check flash address alignment properly.
Do not fill whole padding buffer but its end after original data.
Remove duplicite padding.
Change-Id: Ia5e312909f68d3cc724c8cbffe1cd903b9102124
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/3561
Tested-by: jenkins
Reviewed-by: Steven Stallion <stallion@squareup.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Marc Schink
Paul Fertser
Tomas Vanek
Matthias Welwarsky
Harald Welte
Andrey Smirnov
HarishKumar
Andreas Fritiofson
Rick Foos
Tim Newsome
Jiří Pinkava
Andreas Färber
Gabrielle Trotter
Vivien Didelot
Sean Cross
Sandeep Mistry
Robert Jordens
Michael Dietz
Kent Brinkley
Peter Kuhar
Spencer Oliver
SASANO Takayoshi