You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
kernel_samsung_sm7125/drivers/net/ethernet/intel/i40e/i40e_nvm.c

1517 lines
44 KiB

/*******************************************************************************
*
* Intel Ethernet Controller XL710 Family Linux Driver
* Copyright(c) 2013 - 2014 Intel Corporation.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License along
* with this program. If not, see <http://www.gnu.org/licenses/>.
*
* The full GNU General Public License is included in this distribution in
* the file called "COPYING".
*
* Contact Information:
* e1000-devel Mailing List <e1000-devel@lists.sourceforge.net>
* Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
*
******************************************************************************/
#include "i40e_prototype.h"
/**
* i40e_init_nvm_ops - Initialize NVM function pointers
* @hw: pointer to the HW structure
*
* Setup the function pointers and the NVM info structure. Should be called
* once per NVM initialization, e.g. inside the i40e_init_shared_code().
* Please notice that the NVM term is used here (& in all methods covered
* in this file) as an equivalent of the FLASH part mapped into the SR.
* We are accessing FLASH always thru the Shadow RAM.
**/
i40e_status i40e_init_nvm(struct i40e_hw *hw)
{
struct i40e_nvm_info *nvm = &hw->nvm;
i40e_status ret_code = 0;
u32 fla, gens;
u8 sr_size;
/* The SR size is stored regardless of the nvm programming mode
* as the blank mode may be used in the factory line.
*/
gens = rd32(hw, I40E_GLNVM_GENS);
sr_size = ((gens & I40E_GLNVM_GENS_SR_SIZE_MASK) >>
I40E_GLNVM_GENS_SR_SIZE_SHIFT);
/* Switching to words (sr_size contains power of 2KB) */
nvm->sr_size = BIT(sr_size) * I40E_SR_WORDS_IN_1KB;
/* Check if we are in the normal or blank NVM programming mode */
fla = rd32(hw, I40E_GLNVM_FLA);
if (fla & I40E_GLNVM_FLA_LOCKED_MASK) { /* Normal programming mode */
/* Max NVM timeout */
nvm->timeout = I40E_MAX_NVM_TIMEOUT;
nvm->blank_nvm_mode = false;
} else { /* Blank programming mode */
nvm->blank_nvm_mode = true;
ret_code = I40E_ERR_NVM_BLANK_MODE;
i40e_debug(hw, I40E_DEBUG_NVM, "NVM init error: unsupported blank mode.\n");
}
return ret_code;
}
/**
* i40e_acquire_nvm - Generic request for acquiring the NVM ownership
* @hw: pointer to the HW structure
* @access: NVM access type (read or write)
*
* This function will request NVM ownership for reading
* via the proper Admin Command.
**/
i40e_status i40e_acquire_nvm(struct i40e_hw *hw,
enum i40e_aq_resource_access_type access)
{
i40e_status ret_code = 0;
u64 gtime, timeout;
u64 time_left = 0;
if (hw->nvm.blank_nvm_mode)
goto i40e_i40e_acquire_nvm_exit;
ret_code = i40e_aq_request_resource(hw, I40E_NVM_RESOURCE_ID, access,
0, &time_left, NULL);
/* Reading the Global Device Timer */
gtime = rd32(hw, I40E_GLVFGEN_TIMER);
/* Store the timeout */
hw->nvm.hw_semaphore_timeout = I40E_MS_TO_GTIME(time_left) + gtime;
if (ret_code)
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM acquire type %d failed time_left=%llu ret=%d aq_err=%d\n",
access, time_left, ret_code, hw->aq.asq_last_status);
if (ret_code && time_left) {
/* Poll until the current NVM owner timeouts */
timeout = I40E_MS_TO_GTIME(I40E_MAX_NVM_TIMEOUT) + gtime;
while ((gtime < timeout) && time_left) {
usleep_range(10000, 20000);
gtime = rd32(hw, I40E_GLVFGEN_TIMER);
ret_code = i40e_aq_request_resource(hw,
I40E_NVM_RESOURCE_ID,
access, 0, &time_left,
NULL);
if (!ret_code) {
hw->nvm.hw_semaphore_timeout =
I40E_MS_TO_GTIME(time_left) + gtime;
break;
}
}
if (ret_code) {
hw->nvm.hw_semaphore_timeout = 0;
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM acquire timed out, wait %llu ms before trying again. status=%d aq_err=%d\n",
time_left, ret_code, hw->aq.asq_last_status);
}
}
i40e_i40e_acquire_nvm_exit:
return ret_code;
}
/**
* i40e_release_nvm - Generic request for releasing the NVM ownership
* @hw: pointer to the HW structure
*
* This function will release NVM resource via the proper Admin Command.
**/
void i40e_release_nvm(struct i40e_hw *hw)
{
i40e_status ret_code = I40E_SUCCESS;
u32 total_delay = 0;
if (hw->nvm.blank_nvm_mode)
return;
ret_code = i40e_aq_release_resource(hw, I40E_NVM_RESOURCE_ID, 0, NULL);
/* there are some rare cases when trying to release the resource
* results in an admin Q timeout, so handle them correctly
*/
while ((ret_code == I40E_ERR_ADMIN_QUEUE_TIMEOUT) &&
(total_delay < hw->aq.asq_cmd_timeout)) {
usleep_range(1000, 2000);
ret_code = i40e_aq_release_resource(hw,
I40E_NVM_RESOURCE_ID,
0, NULL);
total_delay++;
}
}
/**
* i40e_poll_sr_srctl_done_bit - Polls the GLNVM_SRCTL done bit
* @hw: pointer to the HW structure
*
* Polls the SRCTL Shadow RAM register done bit.
**/
static i40e_status i40e_poll_sr_srctl_done_bit(struct i40e_hw *hw)
{
i40e_status ret_code = I40E_ERR_TIMEOUT;
u32 srctl, wait_cnt;
/* Poll the I40E_GLNVM_SRCTL until the done bit is set */
for (wait_cnt = 0; wait_cnt < I40E_SRRD_SRCTL_ATTEMPTS; wait_cnt++) {
srctl = rd32(hw, I40E_GLNVM_SRCTL);
if (srctl & I40E_GLNVM_SRCTL_DONE_MASK) {
ret_code = 0;
break;
}
udelay(5);
}
if (ret_code == I40E_ERR_TIMEOUT)
i40e_debug(hw, I40E_DEBUG_NVM, "Done bit in GLNVM_SRCTL not set");
return ret_code;
}
/**
* i40e_read_nvm_word_srctl - Reads Shadow RAM via SRCTL register
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF)
* @data: word read from the Shadow RAM
*
* Reads one 16 bit word from the Shadow RAM using the GLNVM_SRCTL register.
**/
static i40e_status i40e_read_nvm_word_srctl(struct i40e_hw *hw, u16 offset,
u16 *data)
{
i40e_status ret_code = I40E_ERR_TIMEOUT;
u32 sr_reg;
if (offset >= hw->nvm.sr_size) {
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM read error: offset %d beyond Shadow RAM limit %d\n",
offset, hw->nvm.sr_size);
ret_code = I40E_ERR_PARAM;
goto read_nvm_exit;
}
/* Poll the done bit first */
ret_code = i40e_poll_sr_srctl_done_bit(hw);
if (!ret_code) {
/* Write the address and start reading */
sr_reg = ((u32)offset << I40E_GLNVM_SRCTL_ADDR_SHIFT) |
BIT(I40E_GLNVM_SRCTL_START_SHIFT);
wr32(hw, I40E_GLNVM_SRCTL, sr_reg);
/* Poll I40E_GLNVM_SRCTL until the done bit is set */
ret_code = i40e_poll_sr_srctl_done_bit(hw);
if (!ret_code) {
sr_reg = rd32(hw, I40E_GLNVM_SRDATA);
*data = (u16)((sr_reg &
I40E_GLNVM_SRDATA_RDDATA_MASK)
>> I40E_GLNVM_SRDATA_RDDATA_SHIFT);
}
}
if (ret_code)
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM read error: Couldn't access Shadow RAM address: 0x%x\n",
offset);
read_nvm_exit:
return ret_code;
}
/**
* i40e_read_nvm_aq - Read Shadow RAM.
* @hw: pointer to the HW structure.
* @module_pointer: module pointer location in words from the NVM beginning
* @offset: offset in words from module start
* @words: number of words to write
* @data: buffer with words to write to the Shadow RAM
* @last_command: tells the AdminQ that this is the last command
*
* Writes a 16 bit words buffer to the Shadow RAM using the admin command.
**/
static i40e_status i40e_read_nvm_aq(struct i40e_hw *hw, u8 module_pointer,
u32 offset, u16 words, void *data,
bool last_command)
{
i40e_status ret_code = I40E_ERR_NVM;
struct i40e_asq_cmd_details cmd_details;
memset(&cmd_details, 0, sizeof(cmd_details));
cmd_details.wb_desc = &hw->nvm_wb_desc;
/* Here we are checking the SR limit only for the flat memory model.
* We cannot do it for the module-based model, as we did not acquire
* the NVM resource yet (we cannot get the module pointer value).
* Firmware will check the module-based model.
*/
if ((offset + words) > hw->nvm.sr_size)
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM write error: offset %d beyond Shadow RAM limit %d\n",
(offset + words), hw->nvm.sr_size);
else if (words > I40E_SR_SECTOR_SIZE_IN_WORDS)
/* We can write only up to 4KB (one sector), in one AQ write */
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM write fail error: tried to write %d words, limit is %d.\n",
words, I40E_SR_SECTOR_SIZE_IN_WORDS);
else if (((offset + (words - 1)) / I40E_SR_SECTOR_SIZE_IN_WORDS)
!= (offset / I40E_SR_SECTOR_SIZE_IN_WORDS))
/* A single write cannot spread over two sectors */
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM write error: cannot spread over two sectors in a single write offset=%d words=%d\n",
offset, words);
else
ret_code = i40e_aq_read_nvm(hw, module_pointer,
2 * offset, /*bytes*/
2 * words, /*bytes*/
data, last_command, &cmd_details);
return ret_code;
}
/**
* i40e_read_nvm_word_aq - Reads Shadow RAM via AQ
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF)
* @data: word read from the Shadow RAM
*
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
* Reads one 16 bit word from the Shadow RAM using the AdminQ
**/
static i40e_status i40e_read_nvm_word_aq(struct i40e_hw *hw, u16 offset,
u16 *data)
{
i40e_status ret_code = I40E_ERR_TIMEOUT;
ret_code = i40e_read_nvm_aq(hw, 0x0, offset, 1, data, true);
*data = le16_to_cpu(*(__le16 *)data);
return ret_code;
}
/**
* __i40e_read_nvm_word - Reads nvm word, assumes caller does the locking
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF)
* @data: word read from the Shadow RAM
*
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
* Reads one 16 bit word from the Shadow RAM.
*
* Do not use this function except in cases where the nvm lock is already
* taken via i40e_acquire_nvm().
**/
static i40e_status __i40e_read_nvm_word(struct i40e_hw *hw,
u16 offset, u16 *data)
{
i40e_status ret_code = 0;
if (hw->flags & I40E_HW_FLAG_AQ_SRCTL_ACCESS_ENABLE)
ret_code = i40e_read_nvm_word_aq(hw, offset, data);
else
ret_code = i40e_read_nvm_word_srctl(hw, offset, data);
return ret_code;
}
/**
* i40e_read_nvm_word - Reads nvm word and acquire lock if necessary
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF)
* @data: word read from the Shadow RAM
*
* Reads one 16 bit word from the Shadow RAM.
**/
i40e_status i40e_read_nvm_word(struct i40e_hw *hw, u16 offset,
u16 *data)
{
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
i40e_status ret_code = 0;
ret_code = i40e_acquire_nvm(hw, I40E_RESOURCE_READ);
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
if (ret_code)
return ret_code;
ret_code = __i40e_read_nvm_word(hw, offset, data);
i40e_release_nvm(hw);
return ret_code;
}
/**
* i40e_read_nvm_buffer_srctl - Reads Shadow RAM buffer via SRCTL register
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF).
* @words: (in) number of words to read; (out) number of words actually read
* @data: words read from the Shadow RAM
*
* Reads 16 bit words (data buffer) from the SR using the i40e_read_nvm_srrd()
* method. The buffer read is preceded by the NVM ownership take
* and followed by the release.
**/
static i40e_status i40e_read_nvm_buffer_srctl(struct i40e_hw *hw, u16 offset,
u16 *words, u16 *data)
{
i40e_status ret_code = 0;
u16 index, word;
/* Loop thru the selected region */
for (word = 0; word < *words; word++) {
index = offset + word;
ret_code = i40e_read_nvm_word_srctl(hw, index, &data[word]);
if (ret_code)
break;
}
/* Update the number of words read from the Shadow RAM */
*words = word;
return ret_code;
}
/**
* i40e_read_nvm_buffer_aq - Reads Shadow RAM buffer via AQ
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF).
* @words: (in) number of words to read; (out) number of words actually read
* @data: words read from the Shadow RAM
*
* Reads 16 bit words (data buffer) from the SR using the i40e_read_nvm_aq()
* method. The buffer read is preceded by the NVM ownership take
* and followed by the release.
**/
static i40e_status i40e_read_nvm_buffer_aq(struct i40e_hw *hw, u16 offset,
u16 *words, u16 *data)
{
i40e_status ret_code;
u16 read_size = *words;
bool last_cmd = false;
u16 words_read = 0;
u16 i = 0;
do {
/* Calculate number of bytes we should read in this step.
* FVL AQ do not allow to read more than one page at a time or
* to cross page boundaries.
*/
if (offset % I40E_SR_SECTOR_SIZE_IN_WORDS)
read_size = min(*words,
(u16)(I40E_SR_SECTOR_SIZE_IN_WORDS -
(offset % I40E_SR_SECTOR_SIZE_IN_WORDS)));
else
read_size = min((*words - words_read),
I40E_SR_SECTOR_SIZE_IN_WORDS);
/* Check if this is last command, if so set proper flag */
if ((words_read + read_size) >= *words)
last_cmd = true;
ret_code = i40e_read_nvm_aq(hw, 0x0, offset, read_size,
data + words_read, last_cmd);
if (ret_code)
goto read_nvm_buffer_aq_exit;
/* Increment counter for words already read and move offset to
* new read location
*/
words_read += read_size;
offset += read_size;
} while (words_read < *words);
for (i = 0; i < *words; i++)
data[i] = le16_to_cpu(((__le16 *)data)[i]);
read_nvm_buffer_aq_exit:
*words = words_read;
return ret_code;
}
/**
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
* __i40e_read_nvm_buffer - Reads nvm buffer, caller must acquire lock
* @hw: pointer to the HW structure
* @offset: offset of the Shadow RAM word to read (0x000000 - 0x001FFF).
* @words: (in) number of words to read; (out) number of words actually read
* @data: words read from the Shadow RAM
*
* Reads 16 bit words (data buffer) from the SR using the i40e_read_nvm_srrd()
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
* method.
**/
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
static i40e_status __i40e_read_nvm_buffer(struct i40e_hw *hw,
u16 offset, u16 *words,
u16 *data)
{
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
i40e_status ret_code = 0;
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
if (hw->flags & I40E_HW_FLAG_AQ_SRCTL_ACCESS_ENABLE)
ret_code = i40e_read_nvm_buffer_aq(hw, offset, words, data);
else
ret_code = i40e_read_nvm_buffer_srctl(hw, offset, words, data);
return ret_code;
}
/**
* i40e_write_nvm_aq - Writes Shadow RAM.
* @hw: pointer to the HW structure.
* @module_pointer: module pointer location in words from the NVM beginning
* @offset: offset in words from module start
* @words: number of words to write
* @data: buffer with words to write to the Shadow RAM
* @last_command: tells the AdminQ that this is the last command
*
* Writes a 16 bit words buffer to the Shadow RAM using the admin command.
**/
static i40e_status i40e_write_nvm_aq(struct i40e_hw *hw, u8 module_pointer,
u32 offset, u16 words, void *data,
bool last_command)
{
i40e_status ret_code = I40E_ERR_NVM;
struct i40e_asq_cmd_details cmd_details;
memset(&cmd_details, 0, sizeof(cmd_details));
cmd_details.wb_desc = &hw->nvm_wb_desc;
/* Here we are checking the SR limit only for the flat memory model.
* We cannot do it for the module-based model, as we did not acquire
* the NVM resource yet (we cannot get the module pointer value).
* Firmware will check the module-based model.
*/
if ((offset + words) > hw->nvm.sr_size)
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM write error: offset %d beyond Shadow RAM limit %d\n",
(offset + words), hw->nvm.sr_size);
else if (words > I40E_SR_SECTOR_SIZE_IN_WORDS)
/* We can write only up to 4KB (one sector), in one AQ write */
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM write fail error: tried to write %d words, limit is %d.\n",
words, I40E_SR_SECTOR_SIZE_IN_WORDS);
else if (((offset + (words - 1)) / I40E_SR_SECTOR_SIZE_IN_WORDS)
!= (offset / I40E_SR_SECTOR_SIZE_IN_WORDS))
/* A single write cannot spread over two sectors */
i40e_debug(hw, I40E_DEBUG_NVM,
"NVM write error: cannot spread over two sectors in a single write offset=%d words=%d\n",
offset, words);
else
ret_code = i40e_aq_update_nvm(hw, module_pointer,
2 * offset, /*bytes*/
2 * words, /*bytes*/
data, last_command, &cmd_details);
return ret_code;
}
/**
* i40e_calc_nvm_checksum - Calculates and returns the checksum
* @hw: pointer to hardware structure
* @checksum: pointer to the checksum
*
* This function calculates SW Checksum that covers the whole 64kB shadow RAM
* except the VPD and PCIe ALT Auto-load modules. The structure and size of VPD
* is customer specific and unknown. Therefore, this function skips all maximum
* possible size of VPD (1kB).
**/
static i40e_status i40e_calc_nvm_checksum(struct i40e_hw *hw,
u16 *checksum)
{
i40e_status ret_code;
struct i40e_virt_mem vmem;
u16 pcie_alt_module = 0;
u16 checksum_local = 0;
u16 vpd_module = 0;
u16 *data;
u16 i = 0;
ret_code = i40e_allocate_virt_mem(hw, &vmem,
I40E_SR_SECTOR_SIZE_IN_WORDS * sizeof(u16));
if (ret_code)
goto i40e_calc_nvm_checksum_exit;
data = (u16 *)vmem.va;
/* read pointer to VPD area */
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
ret_code = __i40e_read_nvm_word(hw, I40E_SR_VPD_PTR, &vpd_module);
if (ret_code) {
ret_code = I40E_ERR_NVM_CHECKSUM;
goto i40e_calc_nvm_checksum_exit;
}
/* read pointer to PCIe Alt Auto-load module */
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
ret_code = __i40e_read_nvm_word(hw, I40E_SR_PCIE_ALT_AUTO_LOAD_PTR,
&pcie_alt_module);
if (ret_code) {
ret_code = I40E_ERR_NVM_CHECKSUM;
goto i40e_calc_nvm_checksum_exit;
}
/* Calculate SW checksum that covers the whole 64kB shadow RAM
* except the VPD and PCIe ALT Auto-load modules
*/
for (i = 0; i < hw->nvm.sr_size; i++) {
/* Read SR page */
if ((i % I40E_SR_SECTOR_SIZE_IN_WORDS) == 0) {
u16 words = I40E_SR_SECTOR_SIZE_IN_WORDS;
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
ret_code = __i40e_read_nvm_buffer(hw, i, &words, data);
if (ret_code) {
ret_code = I40E_ERR_NVM_CHECKSUM;
goto i40e_calc_nvm_checksum_exit;
}
}
/* Skip Checksum word */
if (i == I40E_SR_SW_CHECKSUM_WORD)
continue;
/* Skip VPD module (convert byte size to word count) */
if ((i >= (u32)vpd_module) &&
(i < ((u32)vpd_module +
(I40E_SR_VPD_MODULE_MAX_SIZE / 2)))) {
continue;
}
/* Skip PCIe ALT module (convert byte size to word count) */
if ((i >= (u32)pcie_alt_module) &&
(i < ((u32)pcie_alt_module +
(I40E_SR_PCIE_ALT_MODULE_MAX_SIZE / 2)))) {
continue;
}
checksum_local += data[i % I40E_SR_SECTOR_SIZE_IN_WORDS];
}
*checksum = (u16)I40E_SR_SW_CHECKSUM_BASE - checksum_local;
i40e_calc_nvm_checksum_exit:
i40e_free_virt_mem(hw, &vmem);
return ret_code;
}
/**
* i40e_update_nvm_checksum - Updates the NVM checksum
* @hw: pointer to hardware structure
*
* NVM ownership must be acquired before calling this function and released
* on ARQ completion event reception by caller.
* This function will commit SR to NVM.
**/
i40e_status i40e_update_nvm_checksum(struct i40e_hw *hw)
{
i40e_status ret_code;
u16 checksum;
__le16 le_sum;
ret_code = i40e_calc_nvm_checksum(hw, &checksum);
if (!ret_code) {
le_sum = cpu_to_le16(checksum);
ret_code = i40e_write_nvm_aq(hw, 0x00, I40E_SR_SW_CHECKSUM_WORD,
1, &le_sum, true);
}
return ret_code;
}
/**
* i40e_validate_nvm_checksum - Validate EEPROM checksum
* @hw: pointer to hardware structure
* @checksum: calculated checksum
*
* Performs checksum calculation and validates the NVM SW checksum. If the
* caller does not need checksum, the value can be NULL.
**/
i40e_status i40e_validate_nvm_checksum(struct i40e_hw *hw,
u16 *checksum)
{
i40e_status ret_code = 0;
u16 checksum_sr = 0;
u16 checksum_local = 0;
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
/* We must acquire the NVM lock in order to correctly synchronize the
* NVM accesses across multiple PFs. Without doing so it is possible
* for one of the PFs to read invalid data potentially indicating that
* the checksum is invalid.
*/
ret_code = i40e_acquire_nvm(hw, I40E_RESOURCE_READ);
if (ret_code)
return ret_code;
ret_code = i40e_calc_nvm_checksum(hw, &checksum_local);
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
__i40e_read_nvm_word(hw, I40E_SR_SW_CHECKSUM_WORD, &checksum_sr);
i40e_release_nvm(hw);
if (ret_code)
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
return ret_code;
/* Verify read checksum from EEPROM is the same as
* calculated checksum
*/
if (checksum_local != checksum_sr)
ret_code = I40E_ERR_NVM_CHECKSUM;
/* If the user cares, return the calculated checksum */
if (checksum)
*checksum = checksum_local;
return ret_code;
}
static i40e_status i40e_nvmupd_state_init(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno);
static i40e_status i40e_nvmupd_state_reading(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno);
static i40e_status i40e_nvmupd_state_writing(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *errno);
static enum i40e_nvmupd_cmd i40e_nvmupd_validate_command(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
int *perrno);
static i40e_status i40e_nvmupd_nvm_erase(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
int *perrno);
static i40e_status i40e_nvmupd_nvm_write(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno);
static i40e_status i40e_nvmupd_nvm_read(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno);
static i40e_status i40e_nvmupd_exec_aq(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno);
static i40e_status i40e_nvmupd_get_aq_result(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno);
static inline u8 i40e_nvmupd_get_module(u32 val)
{
return (u8)(val & I40E_NVM_MOD_PNT_MASK);
}
static inline u8 i40e_nvmupd_get_transaction(u32 val)
{
return (u8)((val & I40E_NVM_TRANS_MASK) >> I40E_NVM_TRANS_SHIFT);
}
static const char * const i40e_nvm_update_state_str[] = {
"I40E_NVMUPD_INVALID",
"I40E_NVMUPD_READ_CON",
"I40E_NVMUPD_READ_SNT",
"I40E_NVMUPD_READ_LCB",
"I40E_NVMUPD_READ_SA",
"I40E_NVMUPD_WRITE_ERA",
"I40E_NVMUPD_WRITE_CON",
"I40E_NVMUPD_WRITE_SNT",
"I40E_NVMUPD_WRITE_LCB",
"I40E_NVMUPD_WRITE_SA",
"I40E_NVMUPD_CSUM_CON",
"I40E_NVMUPD_CSUM_SA",
"I40E_NVMUPD_CSUM_LCB",
"I40E_NVMUPD_STATUS",
"I40E_NVMUPD_EXEC_AQ",
"I40E_NVMUPD_GET_AQ_RESULT",
};
/**
* i40e_nvmupd_command - Process an NVM update command
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* Dispatches command depending on what update state is current
**/
i40e_status i40e_nvmupd_command(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
i40e_status status;
enum i40e_nvmupd_cmd upd_cmd;
/* assume success */
*perrno = 0;
/* early check for status command and debug msgs */
upd_cmd = i40e_nvmupd_validate_command(hw, cmd, perrno);
i40e_debug(hw, I40E_DEBUG_NVM, "%s state %d nvm_release_on_hold %d opc 0x%04x cmd 0x%08x config 0x%08x offset 0x%08x data_size 0x%08x\n",
i40e_nvm_update_state_str[upd_cmd],
hw->nvmupd_state,
hw->nvm_release_on_done, hw->nvm_wait_opcode,
cmd->command, cmd->config, cmd->offset, cmd->data_size);
if (upd_cmd == I40E_NVMUPD_INVALID) {
*perrno = -EFAULT;
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_validate_command returns %d errno %d\n",
upd_cmd, *perrno);
}
/* a status request returns immediately rather than
* going into the state machine
*/
if (upd_cmd == I40E_NVMUPD_STATUS) {
if (!cmd->data_size) {
*perrno = -EFAULT;
return I40E_ERR_BUF_TOO_SHORT;
}
bytes[0] = hw->nvmupd_state;
if (cmd->data_size >= 4) {
bytes[1] = 0;
*((u16 *)&bytes[2]) = hw->nvm_wait_opcode;
}
/* Clear error status on read */
if (hw->nvmupd_state == I40E_NVMUPD_STATE_ERROR)
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
return 0;
}
/* Clear status even it is not read and log */
if (hw->nvmupd_state == I40E_NVMUPD_STATE_ERROR) {
i40e_debug(hw, I40E_DEBUG_NVM,
"Clearing I40E_NVMUPD_STATE_ERROR state without reading\n");
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
}
/* Acquire lock to prevent race condition where adminq_task
* can execute after i40e_nvmupd_nvm_read/write but before state
* variables (nvm_wait_opcode, nvm_release_on_done) are updated.
*
* During NVMUpdate, it is observed that lock could be held for
* ~5ms for most commands. However lock is held for ~60ms for
* NVMUPD_CSUM_LCB command.
*/
mutex_lock(&hw->aq.arq_mutex);
switch (hw->nvmupd_state) {
case I40E_NVMUPD_STATE_INIT:
status = i40e_nvmupd_state_init(hw, cmd, bytes, perrno);
break;
case I40E_NVMUPD_STATE_READING:
status = i40e_nvmupd_state_reading(hw, cmd, bytes, perrno);
break;
case I40E_NVMUPD_STATE_WRITING:
status = i40e_nvmupd_state_writing(hw, cmd, bytes, perrno);
break;
case I40E_NVMUPD_STATE_INIT_WAIT:
case I40E_NVMUPD_STATE_WRITE_WAIT:
/* if we need to stop waiting for an event, clear
* the wait info and return before doing anything else
*/
if (cmd->offset == 0xffff) {
i40e_nvmupd_check_wait_event(hw, hw->nvm_wait_opcode);
status = 0;
goto exit;
}
status = I40E_ERR_NOT_READY;
*perrno = -EBUSY;
break;
default:
/* invalid state, should never happen */
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: no such state %d\n", hw->nvmupd_state);
status = I40E_NOT_SUPPORTED;
*perrno = -ESRCH;
break;
}
exit:
mutex_unlock(&hw->aq.arq_mutex);
return status;
}
/**
* i40e_nvmupd_state_init - Handle NVM update state Init
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* Process legitimate commands of the Init state and conditionally set next
* state. Reject all other commands.
**/
static i40e_status i40e_nvmupd_state_init(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
i40e_status status = 0;
enum i40e_nvmupd_cmd upd_cmd;
upd_cmd = i40e_nvmupd_validate_command(hw, cmd, perrno);
switch (upd_cmd) {
case I40E_NVMUPD_READ_SA:
status = i40e_acquire_nvm(hw, I40E_RESOURCE_READ);
if (status) {
*perrno = i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status);
} else {
status = i40e_nvmupd_nvm_read(hw, cmd, bytes, perrno);
i40e_release_nvm(hw);
}
break;
case I40E_NVMUPD_READ_SNT:
status = i40e_acquire_nvm(hw, I40E_RESOURCE_READ);
if (status) {
*perrno = i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status);
} else {
status = i40e_nvmupd_nvm_read(hw, cmd, bytes, perrno);
if (status)
i40e_release_nvm(hw);
else
hw->nvmupd_state = I40E_NVMUPD_STATE_READING;
}
break;
case I40E_NVMUPD_WRITE_ERA:
status = i40e_acquire_nvm(hw, I40E_RESOURCE_WRITE);
if (status) {
*perrno = i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status);
} else {
status = i40e_nvmupd_nvm_erase(hw, cmd, perrno);
if (status) {
i40e_release_nvm(hw);
} else {
hw->nvm_release_on_done = true;
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_erase;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT_WAIT;
}
}
break;
case I40E_NVMUPD_WRITE_SA:
status = i40e_acquire_nvm(hw, I40E_RESOURCE_WRITE);
if (status) {
*perrno = i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status);
} else {
status = i40e_nvmupd_nvm_write(hw, cmd, bytes, perrno);
if (status) {
i40e_release_nvm(hw);
} else {
hw->nvm_release_on_done = true;
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT_WAIT;
}
}
break;
case I40E_NVMUPD_WRITE_SNT:
status = i40e_acquire_nvm(hw, I40E_RESOURCE_WRITE);
if (status) {
*perrno = i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status);
} else {
status = i40e_nvmupd_nvm_write(hw, cmd, bytes, perrno);
if (status) {
i40e_release_nvm(hw);
} else {
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_WRITE_WAIT;
}
}
break;
case I40E_NVMUPD_CSUM_SA:
status = i40e_acquire_nvm(hw, I40E_RESOURCE_WRITE);
if (status) {
*perrno = i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status);
} else {
status = i40e_update_nvm_checksum(hw);
if (status) {
*perrno = hw->aq.asq_last_status ?
i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status) :
-EIO;
i40e_release_nvm(hw);
} else {
hw->nvm_release_on_done = true;
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT_WAIT;
}
}
break;
case I40E_NVMUPD_EXEC_AQ:
status = i40e_nvmupd_exec_aq(hw, cmd, bytes, perrno);
break;
case I40E_NVMUPD_GET_AQ_RESULT:
status = i40e_nvmupd_get_aq_result(hw, cmd, bytes, perrno);
break;
default:
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: bad cmd %s in init state\n",
i40e_nvm_update_state_str[upd_cmd]);
status = I40E_ERR_NVM;
*perrno = -ESRCH;
break;
}
return status;
}
/**
* i40e_nvmupd_state_reading - Handle NVM update state Reading
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* NVM ownership is already held. Process legitimate commands and set any
* change in state; reject all other commands.
**/
static i40e_status i40e_nvmupd_state_reading(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
i40e_status status = 0;
enum i40e_nvmupd_cmd upd_cmd;
upd_cmd = i40e_nvmupd_validate_command(hw, cmd, perrno);
switch (upd_cmd) {
case I40E_NVMUPD_READ_SA:
case I40E_NVMUPD_READ_CON:
status = i40e_nvmupd_nvm_read(hw, cmd, bytes, perrno);
break;
case I40E_NVMUPD_READ_LCB:
status = i40e_nvmupd_nvm_read(hw, cmd, bytes, perrno);
i40e_release_nvm(hw);
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
break;
default:
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: bad cmd %s in reading state.\n",
i40e_nvm_update_state_str[upd_cmd]);
status = I40E_NOT_SUPPORTED;
*perrno = -ESRCH;
break;
}
return status;
}
/**
* i40e_nvmupd_state_writing - Handle NVM update state Writing
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* NVM ownership is already held. Process legitimate commands and set any
* change in state; reject all other commands
**/
static i40e_status i40e_nvmupd_state_writing(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
i40e_status status = 0;
enum i40e_nvmupd_cmd upd_cmd;
bool retry_attempt = false;
upd_cmd = i40e_nvmupd_validate_command(hw, cmd, perrno);
retry:
switch (upd_cmd) {
case I40E_NVMUPD_WRITE_CON:
status = i40e_nvmupd_nvm_write(hw, cmd, bytes, perrno);
if (!status) {
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_WRITE_WAIT;
}
break;
case I40E_NVMUPD_WRITE_LCB:
status = i40e_nvmupd_nvm_write(hw, cmd, bytes, perrno);
if (status) {
*perrno = hw->aq.asq_last_status ?
i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status) :
-EIO;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
} else {
hw->nvm_release_on_done = true;
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT_WAIT;
}
break;
case I40E_NVMUPD_CSUM_CON:
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
/* Assumes the caller has acquired the nvm */
status = i40e_update_nvm_checksum(hw);
if (status) {
*perrno = hw->aq.asq_last_status ?
i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status) :
-EIO;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
} else {
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_WRITE_WAIT;
}
break;
case I40E_NVMUPD_CSUM_LCB:
i40e: avoid NVM acquire deadlock during NVM update X722 devices use the AdminQ to access the NVM, and this requires taking the AdminQ lock. Because of this, we lock the AdminQ during i40e_read_nvm(), which is also called in places where the lock is already held, such as the firmware update path which wants to lock once and then unlock when finished after performing several tasks. Although this should have only affected X722 devices, commit 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) added locking for all NVM reads, regardless of device family. This resulted in us accidentally causing NVM acquire timeouts on all devices, causing failed firmware updates which left the eeprom in a corrupt state. Create unsafe non-locked variants of i40e_read_nvm_word and i40e_read_nvm_buffer, __i40e_read_nvm_word and __i40e_read_nvm_buffer respectively. These variants will not take the NVM lock and are expected to only be called in places where the NVM lock is already held if needed. Since the only caller of i40e_read_nvm_buffer() was in such a path, remove it entirely in favor of the unsafe version. If necessary we can always add it back in the future. Additionally, we now need to hold the NVM lock in i40e_validate_checksum because the call to i40e_calc_nvm_checksum now assumes that the NVM lock is held. We can further move the call to read I40E_SR_SW_CHECKSUM_WORD up a bit so that we do not need to acquire the NVM lock twice. This should resolve firmware updates and also fix potential raise that could have caused the driver to report an invalid NVM checksum upon driver load. Reported-by: Stefan Assmann <sassmann@kpanic.de> Fixes: 96a39aed25e6 ("i40e: Acquire NVM lock before reads on all devices", 2016-12-02) Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years ago
/* Assumes the caller has acquired the nvm */
status = i40e_update_nvm_checksum(hw);
if (status) {
*perrno = hw->aq.asq_last_status ?
i40e_aq_rc_to_posix(status,
hw->aq.asq_last_status) :
-EIO;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
} else {
hw->nvm_release_on_done = true;
hw->nvm_wait_opcode = i40e_aqc_opc_nvm_update;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT_WAIT;
}
break;
default:
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: bad cmd %s in writing state.\n",
i40e_nvm_update_state_str[upd_cmd]);
status = I40E_NOT_SUPPORTED;
*perrno = -ESRCH;
break;
}
/* In some circumstances, a multi-write transaction takes longer
* than the default 3 minute timeout on the write semaphore. If
* the write failed with an EBUSY status, this is likely the problem,
* so here we try to reacquire the semaphore then retry the write.
* We only do one retry, then give up.
*/
if (status && (hw->aq.asq_last_status == I40E_AQ_RC_EBUSY) &&
!retry_attempt) {
i40e_status old_status = status;
u32 old_asq_status = hw->aq.asq_last_status;
u32 gtime;
gtime = rd32(hw, I40E_GLVFGEN_TIMER);
if (gtime >= hw->nvm.hw_semaphore_timeout) {
i40e_debug(hw, I40E_DEBUG_ALL,
"NVMUPD: write semaphore expired (%d >= %lld), retrying\n",
gtime, hw->nvm.hw_semaphore_timeout);
i40e_release_nvm(hw);
status = i40e_acquire_nvm(hw, I40E_RESOURCE_WRITE);
if (status) {
i40e_debug(hw, I40E_DEBUG_ALL,
"NVMUPD: write semaphore reacquire failed aq_err = %d\n",
hw->aq.asq_last_status);
status = old_status;
hw->aq.asq_last_status = old_asq_status;
} else {
retry_attempt = true;
goto retry;
}
}
}
return status;
}
/**
* i40e_nvmupd_check_wait_event - handle NVM update operation events
* @hw: pointer to the hardware structure
* @opcode: the event that just happened
**/
void i40e_nvmupd_check_wait_event(struct i40e_hw *hw, u16 opcode)
{
if (opcode == hw->nvm_wait_opcode) {
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: clearing wait on opcode 0x%04x\n", opcode);
if (hw->nvm_release_on_done) {
i40e_release_nvm(hw);
hw->nvm_release_on_done = false;
}
hw->nvm_wait_opcode = 0;
if (hw->aq.arq_last_status) {
hw->nvmupd_state = I40E_NVMUPD_STATE_ERROR;
return;
}
switch (hw->nvmupd_state) {
case I40E_NVMUPD_STATE_INIT_WAIT:
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT;
break;
case I40E_NVMUPD_STATE_WRITE_WAIT:
hw->nvmupd_state = I40E_NVMUPD_STATE_WRITING;
break;
default:
break;
}
}
}
/**
* i40e_nvmupd_validate_command - Validate given command
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @perrno: pointer to return error code
*
* Return one of the valid command types or I40E_NVMUPD_INVALID
**/
static enum i40e_nvmupd_cmd i40e_nvmupd_validate_command(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
int *perrno)
{
enum i40e_nvmupd_cmd upd_cmd;
u8 module, transaction;
/* anything that doesn't match a recognized case is an error */
upd_cmd = I40E_NVMUPD_INVALID;
transaction = i40e_nvmupd_get_transaction(cmd->config);
module = i40e_nvmupd_get_module(cmd->config);
/* limits on data size */
if ((cmd->data_size < 1) ||
(cmd->data_size > I40E_NVMUPD_MAX_DATA)) {
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_validate_command data_size %d\n",
cmd->data_size);
*perrno = -EFAULT;
return I40E_NVMUPD_INVALID;
}
switch (cmd->command) {
case I40E_NVM_READ:
switch (transaction) {
case I40E_NVM_CON:
upd_cmd = I40E_NVMUPD_READ_CON;
break;
case I40E_NVM_SNT:
upd_cmd = I40E_NVMUPD_READ_SNT;
break;
case I40E_NVM_LCB:
upd_cmd = I40E_NVMUPD_READ_LCB;
break;
case I40E_NVM_SA:
upd_cmd = I40E_NVMUPD_READ_SA;
break;
case I40E_NVM_EXEC:
if (module == 0xf)
upd_cmd = I40E_NVMUPD_STATUS;
else if (module == 0)
upd_cmd = I40E_NVMUPD_GET_AQ_RESULT;
break;
}
break;
case I40E_NVM_WRITE:
switch (transaction) {
case I40E_NVM_CON:
upd_cmd = I40E_NVMUPD_WRITE_CON;
break;
case I40E_NVM_SNT:
upd_cmd = I40E_NVMUPD_WRITE_SNT;
break;
case I40E_NVM_LCB:
upd_cmd = I40E_NVMUPD_WRITE_LCB;
break;
case I40E_NVM_SA:
upd_cmd = I40E_NVMUPD_WRITE_SA;
break;
case I40E_NVM_ERA:
upd_cmd = I40E_NVMUPD_WRITE_ERA;
break;
case I40E_NVM_CSUM:
upd_cmd = I40E_NVMUPD_CSUM_CON;
break;
case (I40E_NVM_CSUM|I40E_NVM_SA):
upd_cmd = I40E_NVMUPD_CSUM_SA;
break;
case (I40E_NVM_CSUM|I40E_NVM_LCB):
upd_cmd = I40E_NVMUPD_CSUM_LCB;
break;
case I40E_NVM_EXEC:
if (module == 0)
upd_cmd = I40E_NVMUPD_EXEC_AQ;
break;
}
break;
}
return upd_cmd;
}
/**
* i40e_nvmupd_exec_aq - Run an AQ command
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* cmd structure contains identifiers and data buffer
**/
static i40e_status i40e_nvmupd_exec_aq(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
struct i40e_asq_cmd_details cmd_details;
i40e_status status;
struct i40e_aq_desc *aq_desc;
u32 buff_size = 0;
u8 *buff = NULL;
u32 aq_desc_len;
u32 aq_data_len;
i40e_debug(hw, I40E_DEBUG_NVM, "NVMUPD: %s\n", __func__);
memset(&cmd_details, 0, sizeof(cmd_details));
cmd_details.wb_desc = &hw->nvm_wb_desc;
aq_desc_len = sizeof(struct i40e_aq_desc);
memset(&hw->nvm_wb_desc, 0, aq_desc_len);
/* get the aq descriptor */
if (cmd->data_size < aq_desc_len) {
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: not enough aq desc bytes for exec, size %d < %d\n",
cmd->data_size, aq_desc_len);
*perrno = -EINVAL;
return I40E_ERR_PARAM;
}
aq_desc = (struct i40e_aq_desc *)bytes;
/* if data buffer needed, make sure it's ready */
aq_data_len = cmd->data_size - aq_desc_len;
buff_size = max_t(u32, aq_data_len, le16_to_cpu(aq_desc->datalen));
if (buff_size) {
if (!hw->nvm_buff.va) {
status = i40e_allocate_virt_mem(hw, &hw->nvm_buff,
hw->aq.asq_buf_size);
if (status)
i40e_debug(hw, I40E_DEBUG_NVM,
"NVMUPD: i40e_allocate_virt_mem for exec buff failed, %d\n",
status);
}
if (hw->nvm_buff.va) {
buff = hw->nvm_buff.va;
memcpy(buff, &bytes[aq_desc_len], aq_data_len);
}
}
/* and away we go! */
status = i40e_asq_send_command(hw, aq_desc, buff,
buff_size, &cmd_details);
if (status) {
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_exec_aq err %s aq_err %s\n",
i40e_stat_str(hw, status),
i40e_aq_str(hw, hw->aq.asq_last_status));
*perrno = i40e_aq_rc_to_posix(status, hw->aq.asq_last_status);
}
/* should we wait for a followup event? */
if (cmd->offset) {
hw->nvm_wait_opcode = cmd->offset;
hw->nvmupd_state = I40E_NVMUPD_STATE_INIT_WAIT;
}
return status;
}
/**
* i40e_nvmupd_get_aq_result - Get the results from the previous exec_aq
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* cmd structure contains identifiers and data buffer
**/
static i40e_status i40e_nvmupd_get_aq_result(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
u32 aq_total_len;
u32 aq_desc_len;
int remainder;
u8 *buff;
i40e_debug(hw, I40E_DEBUG_NVM, "NVMUPD: %s\n", __func__);
aq_desc_len = sizeof(struct i40e_aq_desc);
aq_total_len = aq_desc_len + le16_to_cpu(hw->nvm_wb_desc.datalen);
/* check offset range */
if (cmd->offset > aq_total_len) {
i40e_debug(hw, I40E_DEBUG_NVM, "%s: offset too big %d > %d\n",
__func__, cmd->offset, aq_total_len);
*perrno = -EINVAL;
return I40E_ERR_PARAM;
}
/* check copylength range */
if (cmd->data_size > (aq_total_len - cmd->offset)) {
int new_len = aq_total_len - cmd->offset;
i40e_debug(hw, I40E_DEBUG_NVM, "%s: copy length %d too big, trimming to %d\n",
__func__, cmd->data_size, new_len);
cmd->data_size = new_len;
}
remainder = cmd->data_size;
if (cmd->offset < aq_desc_len) {
u32 len = aq_desc_len - cmd->offset;
len = min(len, cmd->data_size);
i40e_debug(hw, I40E_DEBUG_NVM, "%s: aq_desc bytes %d to %d\n",
__func__, cmd->offset, cmd->offset + len);
buff = ((u8 *)&hw->nvm_wb_desc) + cmd->offset;
memcpy(bytes, buff, len);
bytes += len;
remainder -= len;
buff = hw->nvm_buff.va;
} else {
buff = hw->nvm_buff.va + (cmd->offset - aq_desc_len);
}
if (remainder > 0) {
int start_byte = buff - (u8 *)hw->nvm_buff.va;
i40e_debug(hw, I40E_DEBUG_NVM, "%s: databuf bytes %d to %d\n",
__func__, start_byte, start_byte + remainder);
memcpy(bytes, buff, remainder);
}
return 0;
}
/**
* i40e_nvmupd_nvm_read - Read NVM
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* cmd structure contains identifiers and data buffer
**/
static i40e_status i40e_nvmupd_nvm_read(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
struct i40e_asq_cmd_details cmd_details;
i40e_status status;
u8 module, transaction;
bool last;
transaction = i40e_nvmupd_get_transaction(cmd->config);
module = i40e_nvmupd_get_module(cmd->config);
last = (transaction == I40E_NVM_LCB) || (transaction == I40E_NVM_SA);
memset(&cmd_details, 0, sizeof(cmd_details));
cmd_details.wb_desc = &hw->nvm_wb_desc;
status = i40e_aq_read_nvm(hw, module, cmd->offset, (u16)cmd->data_size,
bytes, last, &cmd_details);
if (status) {
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_nvm_read mod 0x%x off 0x%x len 0x%x\n",
module, cmd->offset, cmd->data_size);
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_nvm_read status %d aq %d\n",
status, hw->aq.asq_last_status);
*perrno = i40e_aq_rc_to_posix(status, hw->aq.asq_last_status);
}
return status;
}
/**
* i40e_nvmupd_nvm_erase - Erase an NVM module
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @perrno: pointer to return error code
*
* module, offset, data_size and data are in cmd structure
**/
static i40e_status i40e_nvmupd_nvm_erase(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
int *perrno)
{
i40e_status status = 0;
struct i40e_asq_cmd_details cmd_details;
u8 module, transaction;
bool last;
transaction = i40e_nvmupd_get_transaction(cmd->config);
module = i40e_nvmupd_get_module(cmd->config);
last = (transaction & I40E_NVM_LCB);
memset(&cmd_details, 0, sizeof(cmd_details));
cmd_details.wb_desc = &hw->nvm_wb_desc;
status = i40e_aq_erase_nvm(hw, module, cmd->offset, (u16)cmd->data_size,
last, &cmd_details);
if (status) {
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_nvm_erase mod 0x%x off 0x%x len 0x%x\n",
module, cmd->offset, cmd->data_size);
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_nvm_erase status %d aq %d\n",
status, hw->aq.asq_last_status);
*perrno = i40e_aq_rc_to_posix(status, hw->aq.asq_last_status);
}
return status;
}
/**
* i40e_nvmupd_nvm_write - Write NVM
* @hw: pointer to hardware structure
* @cmd: pointer to nvm update command buffer
* @bytes: pointer to the data buffer
* @perrno: pointer to return error code
*
* module, offset, data_size and data are in cmd structure
**/
static i40e_status i40e_nvmupd_nvm_write(struct i40e_hw *hw,
struct i40e_nvm_access *cmd,
u8 *bytes, int *perrno)
{
i40e_status status = 0;
struct i40e_asq_cmd_details cmd_details;
u8 module, transaction;
bool last;
transaction = i40e_nvmupd_get_transaction(cmd->config);
module = i40e_nvmupd_get_module(cmd->config);
last = (transaction & I40E_NVM_LCB);
memset(&cmd_details, 0, sizeof(cmd_details));
cmd_details.wb_desc = &hw->nvm_wb_desc;
status = i40e_aq_update_nvm(hw, module, cmd->offset,
(u16)cmd->data_size, bytes, last,
&cmd_details);
if (status) {
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_nvm_write mod 0x%x off 0x%x len 0x%x\n",
module, cmd->offset, cmd->data_size);
i40e_debug(hw, I40E_DEBUG_NVM,
"i40e_nvmupd_nvm_write status %d aq %d\n",
status, hw->aq.asq_last_status);
*perrno = i40e_aq_rc_to_posix(status, hw->aq.asq_last_status);
}
return status;
}