* [PATCH 2/3] imx9: scmi: Correct shutdown checking
2026-06-18 1:17 [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR ye.li
@ 2026-06-18 1:17 ` ye.li
2026-06-18 1:17 ` [PATCH 3/3] imx9: scmi: use 48-bit MAC add instead of incrementing mac[5] ye.li
` (2 subsequent siblings)
3 siblings, 0 replies; 8+ messages in thread
From: ye.li @ 2026-06-18 1:17 UTC (permalink / raw)
To: festevam, u-boot; +Cc: peng.fan, uboot-imx, alice.guo, ye.li
From: Peng Fan <peng.fan@nxp.com>
The shutdown checking should use shutdown flags, not bootflags.
Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
---
arch/arm/mach-imx/imx9/scmi/soc.c | 20 ++++++++++----------
1 file changed, 10 insertions(+), 10 deletions(-)
diff --git a/arch/arm/mach-imx/imx9/scmi/soc.c b/arch/arm/mach-imx/imx9/scmi/soc.c
index ce86cdf8574..4341a061d08 100644
--- a/arch/arm/mach-imx/imx9/scmi/soc.c
+++ b/arch/arm/mach-imx/imx9/scmi/soc.c
@@ -741,11 +741,11 @@ int get_reset_reason(bool sys, bool lm)
}
if (out.shutdownflags & MISC_SHUTDOWN_FLAG_VLD) {
printf("SYS shutdown reason: %s, origin: %ld, errid: %ld\n",
- rst[out.bootflags & MISC_SHUTDOWN_FLAG_REASON],
- out.bootflags & MISC_SHUTDOWN_FLAG_ORG_VLD ?
- FIELD_GET(MISC_SHUTDOWN_FLAG_ORIGIN, out.bootflags) : -1,
- out.bootflags & MISC_SHUTDOWN_FLAG_ERR_VLD ?
- FIELD_GET(MISC_SHUTDOWN_FLAG_ERR_ID, out.bootflags) : -1
+ rst[out.shutdownflags & MISC_SHUTDOWN_FLAG_REASON],
+ out.shutdownflags & MISC_SHUTDOWN_FLAG_ORG_VLD ?
+ FIELD_GET(MISC_SHUTDOWN_FLAG_ORIGIN, out.shutdownflags) : -1,
+ out.shutdownflags & MISC_SHUTDOWN_FLAG_ERR_VLD ?
+ FIELD_GET(MISC_SHUTDOWN_FLAG_ERR_ID, out.shutdownflags) : -1
);
}
}
@@ -772,11 +772,11 @@ int get_reset_reason(bool sys, bool lm)
if (out.shutdownflags & MISC_SHUTDOWN_FLAG_VLD) {
printf("LM shutdown reason: %s, origin: %ld, errid: %ld\n",
- rst[out.bootflags & MISC_SHUTDOWN_FLAG_REASON],
- out.bootflags & MISC_SHUTDOWN_FLAG_ORG_VLD ?
- FIELD_GET(MISC_SHUTDOWN_FLAG_ORIGIN, out.bootflags) : -1,
- out.bootflags & MISC_SHUTDOWN_FLAG_ERR_VLD ?
- FIELD_GET(MISC_SHUTDOWN_FLAG_ERR_ID, out.bootflags) : -1
+ rst[out.shutdownflags & MISC_SHUTDOWN_FLAG_REASON],
+ out.shutdownflags & MISC_SHUTDOWN_FLAG_ORG_VLD ?
+ FIELD_GET(MISC_SHUTDOWN_FLAG_ORIGIN, out.shutdownflags) : -1,
+ out.shutdownflags & MISC_SHUTDOWN_FLAG_ERR_VLD ?
+ FIELD_GET(MISC_SHUTDOWN_FLAG_ERR_ID, out.shutdownflags) : -1
);
}
}
--
2.34.1
^ permalink raw reply related [flat|nested] 8+ messages in thread* [PATCH 3/3] imx9: scmi: use 48-bit MAC add instead of incrementing mac[5]
2026-06-18 1:17 [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR ye.li
2026-06-18 1:17 ` [PATCH 2/3] imx9: scmi: Correct shutdown checking ye.li
@ 2026-06-18 1:17 ` ye.li
2026-06-18 2:03 ` Peng Fan
2026-06-18 2:02 ` [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR Peng Fan
2026-06-18 7:00 ` Francesco Dolcini
3 siblings, 1 reply; 8+ messages in thread
From: ye.li @ 2026-06-18 1:17 UTC (permalink / raw)
To: festevam, u-boot; +Cc: peng.fan, uboot-imx, alice.guo, ye.li
From: Alice Guo <alice.guo@nxp.com>
Replace mac[5] += offset with a proper 48-bit MAC address add to avoid
overflow issues when mac[5] was close to 255.
Example:
i.MX95 MAC2: 00:11:22:33:44:FE + 6
Old: 00:11:22:33:44:04 (overflow, wraps to 0)
New: 00:11:22:33:45:04 (correct carry from mac[5] to mac[4])
Signed-off-by: Alice Guo <alice.guo@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>
---
arch/arm/mach-imx/imx9/scmi/soc.c | 39 +++++++++++++++++++++++++++----
1 file changed, 34 insertions(+), 5 deletions(-)
diff --git a/arch/arm/mach-imx/imx9/scmi/soc.c b/arch/arm/mach-imx/imx9/scmi/soc.c
index 4341a061d08..5a635ba8433 100644
--- a/arch/arm/mach-imx/imx9/scmi/soc.c
+++ b/arch/arm/mach-imx/imx9/scmi/soc.c
@@ -583,6 +583,35 @@ phys_size_t get_effective_memsize(void)
}
}
+static inline u64 ether_addr_to_u64(const u8 *addr)
+{
+ u64 u = 0;
+ int i;
+
+ for (i = 0; i < 6; i++)
+ u = u << 8 | addr[i];
+
+ return u;
+}
+
+static inline void u64_to_ether_addr(u64 u, u8 *addr)
+{
+ int i;
+
+ for (i = 6 - 1; i >= 0; i--) {
+ addr[i] = u & 0xff;
+ u = u >> 8;
+ }
+}
+
+static inline void eth_addr_add(u8 *addr, long offset)
+{
+ u64 u = ether_addr_to_u64(addr);
+
+ u += offset;
+ u64_to_ether_addr(u, addr);
+}
+
void imx_get_mac_from_fuse(int dev_id, unsigned char *mac)
{
u32 val[2] = {};
@@ -627,16 +656,16 @@ void imx_get_mac_from_fuse(int dev_id, unsigned char *mac)
* | 10 | netc switch | swp2 |
*/
if (dev_id == 0)
- mac[5] = mac[5] + 2; /* enetc3 mac/swp0 */
+ eth_addr_add(mac, 2); /* enetc3 mac/swp0 */
if (dev_id == 1)
- mac[5] = mac[5] + 8; /* enetc1 */
+ eth_addr_add(mac, 8); /* enetc1 */
if (dev_id == 2)
- mac[5] = mac[5] + 9; /* enetc2 */
+ eth_addr_add(mac, 9); /* enetc2 */
} else {
if (dev_id == 1)
- mac[5] = mac[5] + 3;
+ eth_addr_add(mac, 3);
if (dev_id == 2)
- mac[5] = mac[5] + 6;
+ eth_addr_add(mac, 6);
}
debug("%s: MAC%d: %pM\n", __func__, dev_id, mac);
--
2.34.1
^ permalink raw reply related [flat|nested] 8+ messages in thread* Re: [PATCH 3/3] imx9: scmi: use 48-bit MAC add instead of incrementing mac[5]
2026-06-18 1:17 ` [PATCH 3/3] imx9: scmi: use 48-bit MAC add instead of incrementing mac[5] ye.li
@ 2026-06-18 2:03 ` Peng Fan
0 siblings, 0 replies; 8+ messages in thread
From: Peng Fan @ 2026-06-18 2:03 UTC (permalink / raw)
To: ye.li; +Cc: festevam, u-boot, peng.fan, uboot-imx, alice.guo, ye.li
On Thu, Jun 18, 2026 at 09:17:49AM +0800, ye.li@oss.nxp.com wrote:
>From: Alice Guo <alice.guo@nxp.com>
>
>Replace mac[5] += offset with a proper 48-bit MAC address add to avoid
>overflow issues when mac[5] was close to 255.
>
>Example:
>i.MX95 MAC2: 00:11:22:33:44:FE + 6
>Old: 00:11:22:33:44:04 (overflow, wraps to 0)
>New: 00:11:22:33:45:04 (correct carry from mac[5] to mac[4])
>
>Signed-off-by: Alice Guo <alice.guo@nxp.com>
>Signed-off-by: Ye Li <ye.li@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR
2026-06-18 1:17 [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR ye.li
2026-06-18 1:17 ` [PATCH 2/3] imx9: scmi: Correct shutdown checking ye.li
2026-06-18 1:17 ` [PATCH 3/3] imx9: scmi: use 48-bit MAC add instead of incrementing mac[5] ye.li
@ 2026-06-18 2:02 ` Peng Fan
2026-06-18 7:00 ` Francesco Dolcini
3 siblings, 0 replies; 8+ messages in thread
From: Peng Fan @ 2026-06-18 2:02 UTC (permalink / raw)
To: ye.li; +Cc: festevam, u-boot, peng.fan, uboot-imx, alice.guo, ye.li
On Thu, Jun 18, 2026 at 09:17:47AM +0800, ye.li@oss.nxp.com wrote:
>From: Ye Li <ye.li@nxp.com>
>
>After supporting get DRAM size from SM, the trampoline buffer address
>still depends on PHYS_SDRAM_SIZE. If the real DDR size is less than
>PHYS_SDRAM_SIZE, the trampoline buffer address is invalid and SPL will
>crash. So use board_phys_sdram_size to get real DDR size to calculate
>correct address.
>
>Fixes: e1cc7117b630 ("imx9: scmi: Get DDR size through SM SCMI API")
>Signed-off-by: Ye Li <ye.li@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
^ permalink raw reply [flat|nested] 8+ messages in thread* Re: [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR
2026-06-18 1:17 [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR ye.li
` (2 preceding siblings ...)
2026-06-18 2:02 ` [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR Peng Fan
@ 2026-06-18 7:00 ` Francesco Dolcini
2026-06-18 8:14 ` Ye Li
3 siblings, 1 reply; 8+ messages in thread
From: Francesco Dolcini @ 2026-06-18 7:00 UTC (permalink / raw)
To: ye.li, Emanuele Ghidoli
Cc: festevam, u-boot, peng.fan, uboot-imx, alice.guo, ye.li
+ Emanuele
On Thu, Jun 18, 2026 at 09:17:47AM +0800, ye.li@oss.nxp.com wrote:
> From: Ye Li <ye.li@nxp.com>
>
> After supporting get DRAM size from SM, the trampoline buffer address
> still depends on PHYS_SDRAM_SIZE. If the real DDR size is less than
> PHYS_SDRAM_SIZE, the trampoline buffer address is invalid and SPL will
> crash. So use board_phys_sdram_size to get real DDR size to calculate
> correct address.
>
> Fixes: e1cc7117b630 ("imx9: scmi: Get DDR size through SM SCMI API")
> Signed-off-by: Ye Li <ye.li@nxp.com>
> ---
> arch/arm/mach-imx/imx9/scmi/soc.c | 16 +++++++++++-----
> 1 file changed, 11 insertions(+), 5 deletions(-)
>
> diff --git a/arch/arm/mach-imx/imx9/scmi/soc.c b/arch/arm/mach-imx/imx9/scmi/soc.c
> index 00b8693fbe0..ce86cdf8574 100644
> --- a/arch/arm/mach-imx/imx9/scmi/soc.c
> +++ b/arch/arm/mach-imx/imx9/scmi/soc.c
> @@ -1177,10 +1177,11 @@ enum boot_device get_boot_device(void)
>
> bool arch_check_dst_in_secure(void *start, ulong size)
> {
> - ulong ns_end = CFG_SYS_SDRAM_BASE + PHYS_SDRAM_SIZE;
> -#ifdef PHYS_SDRAM_2_SIZE
> - ns_end += PHYS_SDRAM_2_SIZE;
> -#endif
> + ulong ns_end;
> + phys_size_t dram_size;
> +
> + board_phys_sdram_size(&dram_size);
Emanuele: does this change has any impact on the memory size detection we have
implemented in our i.MX95 based boards?
Peng: board_phys_sdram_size() in board/toradex/verdin-imx95/verdin-imx95.c
Francesco
^ permalink raw reply [flat|nested] 8+ messages in thread* Re: [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR
2026-06-18 7:00 ` Francesco Dolcini
@ 2026-06-18 8:14 ` Ye Li
2026-06-18 8:41 ` Emanuele Ghidoli
0 siblings, 1 reply; 8+ messages in thread
From: Ye Li @ 2026-06-18 8:14 UTC (permalink / raw)
To: Francesco Dolcini, Emanuele Ghidoli
Cc: festevam, u-boot, peng.fan, uboot-imx, alice.guo, ye.li
On 6/18/2026 3:00 PM, Francesco Dolcini wrote:
> + Emanuele
>
> On Thu, Jun 18, 2026 at 09:17:47AM +0800, ye.li@oss.nxp.com wrote:
>> From: Ye Li <ye.li@nxp.com>
>>
>> After supporting get DRAM size from SM, the trampoline buffer address
>> still depends on PHYS_SDRAM_SIZE. If the real DDR size is less than
>> PHYS_SDRAM_SIZE, the trampoline buffer address is invalid and SPL will
>> crash. So use board_phys_sdram_size to get real DDR size to calculate
>> correct address.
>>
>> Fixes: e1cc7117b630 ("imx9: scmi: Get DDR size through SM SCMI API")
>> Signed-off-by: Ye Li <ye.li@nxp.com>
>> ---
>> arch/arm/mach-imx/imx9/scmi/soc.c | 16 +++++++++++-----
>> 1 file changed, 11 insertions(+), 5 deletions(-)
>>
>> diff --git a/arch/arm/mach-imx/imx9/scmi/soc.c b/arch/arm/mach-imx/imx9/scmi/soc.c
>> index 00b8693fbe0..ce86cdf8574 100644
>> --- a/arch/arm/mach-imx/imx9/scmi/soc.c
>> +++ b/arch/arm/mach-imx/imx9/scmi/soc.c
>> @@ -1177,10 +1177,11 @@ enum boot_device get_boot_device(void)
>>
>> bool arch_check_dst_in_secure(void *start, ulong size)
>> {
>> - ulong ns_end = CFG_SYS_SDRAM_BASE + PHYS_SDRAM_SIZE;
>> -#ifdef PHYS_SDRAM_2_SIZE
>> - ns_end += PHYS_SDRAM_2_SIZE;
>> -#endif
>> + ulong ns_end;
>> + phys_size_t dram_size;
>> +
>> + board_phys_sdram_size(&dram_size);
>
> Emanuele: does this change has any impact on the memory size detection we have
> implemented in our i.MX95 based boards?
>
> Peng: board_phys_sdram_size() in board/toradex/verdin-imx95/verdin-imx95.c
>
It won't have impact on the memory size detection. It uses the output of
memory size detection for selection correct address for trampoline buffer.
Previously trampoline buffer uses PHYS_SDRAM_SIZE which is defined as
2GB - 256MB, if the real DDR size detected at runtime is less than
PHYS_SDRAM_SIZE (for example 1GB), this buffer address is invalid.
Best regards,
Ye Li
> Francesco
>
^ permalink raw reply [flat|nested] 8+ messages in thread* Re: [PATCH 1/3] imx9: scmi: Fix SPL trampoline buffer for 1GB DDR
2026-06-18 8:14 ` Ye Li
@ 2026-06-18 8:41 ` Emanuele Ghidoli
0 siblings, 0 replies; 8+ messages in thread
From: Emanuele Ghidoli @ 2026-06-18 8:41 UTC (permalink / raw)
To: Ye Li, Francesco Dolcini, Emanuele Ghidoli
Cc: festevam, u-boot, peng.fan, uboot-imx, alice.guo, ye.li
On 6/18/26 10:14, Ye Li wrote:
>
>
> On 6/18/2026 3:00 PM, Francesco Dolcini wrote:
>> + Emanuele
>>
>> On Thu, Jun 18, 2026 at 09:17:47AM +0800, ye.li@oss.nxp.com wrote:
>>> From: Ye Li <ye.li@nxp.com>
>>>
>>> After supporting get DRAM size from SM, the trampoline buffer address
>>> still depends on PHYS_SDRAM_SIZE. If the real DDR size is less than
>>> PHYS_SDRAM_SIZE, the trampoline buffer address is invalid and SPL will
>>> crash. So use board_phys_sdram_size to get real DDR size to calculate
>>> correct address.
>>>
>>> Fixes: e1cc7117b630 ("imx9: scmi: Get DDR size through SM SCMI API")
>>> Signed-off-by: Ye Li <ye.li@nxp.com>
>>> ---
>>> arch/arm/mach-imx/imx9/scmi/soc.c | 16 +++++++++++-----
>>> 1 file changed, 11 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/arch/arm/mach-imx/imx9/scmi/soc.c b/arch/arm/mach-imx/imx9/
>>> scmi/soc.c
>>> index 00b8693fbe0..ce86cdf8574 100644
>>> --- a/arch/arm/mach-imx/imx9/scmi/soc.c
>>> +++ b/arch/arm/mach-imx/imx9/scmi/soc.c
>>> @@ -1177,10 +1177,11 @@ enum boot_device get_boot_device(void)
>>> bool arch_check_dst_in_secure(void *start, ulong size)
>>> {
>>> - ulong ns_end = CFG_SYS_SDRAM_BASE + PHYS_SDRAM_SIZE;
>>> -#ifdef PHYS_SDRAM_2_SIZE
>>> - ns_end += PHYS_SDRAM_2_SIZE;
>>> -#endif
>>> + ulong ns_end;
>>> + phys_size_t dram_size;
>>> +
>>> + board_phys_sdram_size(&dram_size);
>>
>> Emanuele: does this change has any impact on the memory size detection we have
>> implemented in our i.MX95 based boards?
>>
>> Peng: board_phys_sdram_size() in board/toradex/verdin-imx95/verdin-imx95.c
>>
> It won't have impact on the memory size detection. It uses the output of
> memory size detection for selection correct address for trampoline buffer.
> Previously trampoline buffer uses PHYS_SDRAM_SIZE which is defined as 2GB -
> 256MB, if the real DDR size detected at runtime is less than PHYS_SDRAM_SIZE
> (for example 1GB), this buffer address is invalid.
>
> Best regards,
> Ye Li
>
>> Francesco
>>
>
While it is not obvious that board_phys_sdram_size() is safe in all contexts,
it can be called here without side effects.
Tested-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
^ permalink raw reply [flat|nested] 8+ messages in thread