Cold-booted latest snapshot and, as @gmtii says, doesn't work.
I also went through the patch, and I couldn't find anything obvious.
Unfortunately, standard OpenWRT images since mkrle's patch was integrated have an issue enabling the modem, and we haven't been able to find a fix for that.
For what it's worth, I've applied @mkrle patches on 2025/10/31 version, and modem is successfully detected.
Best regards.
Please, can you post the OpenWRT version plus the @mkrle patches, or the patched firmware? Can it be applied via ssh? Thank you.
Did anyone check, if this openwrt image would work on HH500E variant ? Apparently, soc is the same, only i canāt find anything about RAM or nand flash ?
Based on my knowledge it hasn't been tested but it should work. But don't try it without a serial console.
Hi,
I've narrowed the search and the suspect commit is around 11/20.
Maybe the kernel bump from 6.12 to 6.12.58, and the only qualcommax commit is the culprit?
Best regards.
Yes, jav has already confirmed itās the kernel bump that broke it. I donāt think itās the qcom NSS patch because thatās related to network ports while the modem is on PCI/MHI bus. But I am not 100% certain.
The elephant in the room is the MHI PM patch, but looking at the code I donāt see how that could break it. There are also a few PCI patches including one for PM, these is where I suspect the problem could be.
My mistake for not reading @jav's post properly 
. Which MHI patch are you referring to?
Regards.
This one:
Tried to revert this patch on the latest snapshot (kernel 6.12.74) and it doesn't fix the problem.
I guess for now we need to hope that https://github.com/openwrt/openwrt/pull/22276 will fix it.
Sadly, it doesn't fix it either.
Linux OpenWrt 6.12.77 #0 SMP Sat Mar 14 14:33:30 2026 aarch64 GNU/Linux
Regads.
Hi,
I think this is the patch that's causing the problem:
Regards.
It is! Latest snapshot (kernel 6.12.74), with this patch reverted, works (after a power cycle).
Great!!! 
Hi,
I suppose we can send a patch to revert this commit @mkrle .
Regards.
From:
Date: Thu, 19 Mar 2026 00:00:00 +0100
Subject: [PATCH] Revert "irqchip/gic-v2m: Handle Multiple MSI base IRQ Alignment"
This reverts upstream commit 2ef3886ce626dcdab0cbc452dbbebc19f57133d8.
On this platform, upgrading from Linux 6.12.57 to 6.12.58 breaks SDX55
operation over MHI. The modem reaches AMSS/M0/READY, but channel START/RESET
command completions fail. Reverting this change restores the previous GICv2m
MSI allocation behavior for multi-MSI requests.
--- a/drivers/irqchip/irq-gic-v2m.c
+++ b/drivers/irqchip/irq-gic-v2m.c
@@ -156,17 +156,13 @@
{
msi_alloc_info_t *info = args;
struct v2m_data *v2m = NULL, *tmp;
- int hwirq, i, err = 0;
- unsigned long offset;
- unsigned long align_mask = nr_irqs - 1;
+ int hwirq, offset, i, err 0;
spin_lock(&v2m_lock);
list_for_each_entry(tmp, &v2m_nodes, entry) {
- unsigned long align_off = tmp->spi_start - (tmp->spi_start & ~align_mask);
-
- offset = bitmap_find_next_zero_area_off(tmp->bm, tmp->nr_spis, 0,
- nr_irqs, align_mask, align_off);
- if (offset < tmp->nr_spis) {
+ offset = bitmap_find_free_region(tmp->bm, tmp->nr_spis,
+ get_count_order(nr_irqs));
+ if (offset >= 0) {
v2m = tmp;
bitmap_set(v2m->bm, offset, nr_irqs);
break;
I guess the original patch fixed something, didn't it? I haven't found the rational behind it to understand why it was needed.
Just to make a disclaimer, things this low level are beyond my skills. But I can still guess.
I haven't found any complaints about this elsewhere on the internet, and the patch has trickled down to stable kernel. I think it could have something to do with this but I'm not sure:
We diverge from upstream here, but this patch has been in place for at least 3 years.
You can submit a PR as per your suggestion (and you should if we don't find anything else), but based on my experience it may not be easy to have it accepted.
I think only @robimarko can help us here, as any PR will likely end up with him anyway, not sure if he reads the forums. If he does, then just to summarize, this patch has broken the 5G PCI/MHI modem for our device, and we're looking for ways to fix it.