Skip to contentSkip to conversation

various memory related fixups

PATCH Superseded [email protected] 4 participants 12 messages May 13 – Jun 3

Subsystems (2)

v8 of 10 Superseded Latest is v10 →

lore
Share & tools
lore
#0 lore raw

v8 cover various memory related fixups

From: Randolph Sapp <[email protected]>

Nitpicks and fixes from the discovery thread on adding PocketBeagle2 support [1]. This does a lot of general setup required for the device, but these modifications themselves aren't device specific. For those specifically interested in PocketBeagle2 support and don't care about these details, my development branch is public [2].

That first patch may provoke some opinions, but honestly if that warning was still present I wouldn't have spent a week poking holes in both the EFI and LMB allocations systems. Please let me know if there is a specific usecase that it breaks though.

[1] https://lore.kernel.org/all/[email protected]/
[2] https://github.com/StaticRocket/u-boot/tree/feature/pocketbeagle2

v2:
- Remove additional increment and decrement in lmb_free_fdt_regions
- Drop the patch to backfill EFI_CONVENTIONAL_MEMORY
- Adjust the removal loop nitpick patch description
- Change the reserve memory patch to use new end_addr_sp
v3:
- Update lmb flags to use the macro documentation for constants
- Change efi_mem_sort to use list_for_each_entry_safe
v4:
- Fix typos in LMB allocation flags macro documentation
- Rename end_addr_sp to initial_relocaddr
- Keep the map_sysmem dance in the efi u-boot reservation
- Use the active device tree pointed to by gd->fdt_blob to clean up old reservations
v5:
- Keep return value as long in boot_fdt_reserve_region
- Fix formatting in initial_relocaddr patch
v6:
- Drop patches that have been picked up already
- Add board_get_usable_ram_top for the sandbox
- s/boot_fdt_reserve_region/boot_fdt_handle_region/
v7:
- Remove board_get_usable_ram_top for the sandbox
- Reinstate bank hopping logic for U-Boot reserved region
- Update description for gd->initial_relocaddr, make it clear this is also an exclusive value
- Add disclaimer to boot_fdt_add_mem_rsv_regions
- Add static boot_fdt_handle_mem_rsv_regions as a generic walker for the two calls into boot_fdt_add_mem_rsv_regions
- Make boot_fdt_add_mem_rsv_regions fdt pointer const
- Leave the PRAM region out of the reservation. Previous commit messages indicate that this is intended.
- Add a test for boot_fdt_add_mem_rsv_regions
v8:
- Adjust initial_relocaddr doc string
- Convert ref to fixes tag as requested
- Rework test_boot_fdt_add_mem_rsv_regions to mitigate lingering state

Randolph Sapp (3):
boot: image-fdt: free old dtb reservations
test: boot: add a fdt reserved region check
memory: reserve from start_addr_sp to initial_relocaddr

 boot/image-fdt.c                  | 77 +++++++++++++++++++++++--------
 common/board_f.c                  |  9 +++-
 include/asm-generic/global_data.h |  9 ++++
 include/image.h                   |  2 +-
 lib/efi_loader/efi_memory.c       |  2 +-
 lib/lmb.c                         |  7 +--
 test/boot/Makefile                |  1 +
 test/boot/image_fdt.c             | 70 ++++++++++++++++++++++++++++
 test/cmd_ut.c                     |  2 +
 test/py/tests/test_suite.py       |  2 +-
 10 files changed, 154 insertions(+), 27 deletions(-)
 create mode 100644 test/boot/image_fdt.c
Signature · 1 line
2.54.0
#1 lore raw

PATCH v8 1/3 boot: image-fdt: free old dtb reservations

From: Randolph Sapp <[email protected]>

Add a free flag and an initial call to free allocations covered by the global FDT. This assumes that all calls to boot_fdt_add_mem_rsv_regions occur before the transition to the new device tree, thus we can access the currently active device tree through the global data pointer.

This allows us to clearly indicate to the user when a device tree reservation fails. How we handle this can still use some improvement. Right now we'll keep the default behavior and try to boot anyway.

Fixes 5a6aa7d5913 ("boot: fdt: Handle already reserved memory in boot_fdt_reserve_region()")

---

v7:
- Add disclaimer to boot_fdt_add_mem_rsv_regions
- Add static boot_fdt_handle_mem_rsv_regions as a generic walker for the two calls into boot_fdt_add_mem_rsv_regions
- Make boot_fdt_add_mem_rsv_regions fdt pointer const
v8:
- Convert ref to fixes tag as requested

 boot/image-fdt.c | 77 +++++++++++++++++++++++++++++++++++-------------
 include/image.h  |  2 +-
 2 files changed, 58 insertions(+), 21 deletions(-)
boot/image-fdt.c +57 -20
index a3a4fb8b558..1150131a11e 100644
@@ -69,35 +69,50 @@ static const struct legacy_img_hdr *image_get_fdt(ulong fdt_addr)
 }
 #endif
 
-static void boot_fdt_reserve_region(u64 addr, u64 size, u32 flags)
+/**
+ * boot_fdt_handle_region - Reserve or free a given FDT region in LMB
+ * @addr: Reservation base address
+ * @size: Reservation size
+ * @flags: Reservation flags
+ * @free: Indicate if region is being freed or allocated
+ *
+ * Add or free a given reservation from LMB. This reports to the user if any
+ * errors occurred during either operation.
+ */
+static void boot_fdt_handle_region(u64 addr, u64 size, u32 flags, bool free)
 {
 	long ret;
 	phys_addr_t rsv_addr;
 
 	rsv_addr = (phys_addr_t)addr;
-	ret = lmb_alloc_mem(LMB_MEM_ALLOC_ADDR, 0, &rsv_addr, size, flags);
+	if (free)
+		ret = lmb_free(rsv_addr, size, flags);
+	else
+		ret = lmb_alloc_mem(LMB_MEM_ALLOC_ADDR, 0, &rsv_addr, size,
+				    flags);
+
 	if (!ret) {
-		debug("   reserving fdt memory region: addr=%llx size=%llx flags=%x\n",
-		      (unsigned long long)addr,
+		debug("   %s fdt memory region: addr=%llx size=%llx flags=%x\n",
+		      free ? "freed" : "reserved", (unsigned long long)addr,
 		      (unsigned long long)size, flags);
-	} else if (ret != -EEXIST && ret != -EINVAL) {
-		puts("ERROR: reserving fdt memory region failed ");
-		printf("(addr=%llx size=%llx flags=%x)\n",
-		       (unsigned long long)addr,
-		       (unsigned long long)size, flags);
+	} else {
+		printf("ERROR: %s fdt memory region failed (addr=%llx size=%llx flags=%x): %ld\n",
+		       free ? "freeing" : "reserving", (unsigned long long)addr,
+		       (unsigned long long)size, flags, ret);
 	}
 }
 
 /**
- * boot_fdt_add_mem_rsv_regions - Mark the memreserve and reserved-memory
- * sections as unusable
+ * boot_fdt_handle_mem_rsv_regions - Handle FDT memreserve and reserved-memory
+ *                                   sections
  * @fdt_blob: pointer to fdt blob base address
+ * @free: indicate if regions are being freed
  *
- * Adds the and reserved-memorymemreserve regions in the dtb to the lmb block.
- * Adding the memreserve regions prevents u-boot from using them to store the
- * initrd or the fdt blob.
+ * Adds or removes reserved-memory and memreserve regions in the dtb to the lmb
+ * block. Adding the memreserve regions prevents u-boot from using them to store
+ * the initrd or the fdt blob.
  */
-void boot_fdt_add_mem_rsv_regions(void *fdt_blob)
+static void boot_fdt_handle_mem_rsv_regions(const void *fdt_blob, bool free)
 {
 	uint64_t addr, size;
 	int i, total, ret;
@@ -105,15 +120,12 @@ void boot_fdt_add_mem_rsv_regions(void *fdt_blob)
 	struct fdt_resource res;
 	u32 flags;
 
-	if (fdt_check_header(fdt_blob) != 0)
-		return;
-
 	/* process memreserve sections */
 	total = fdt_num_mem_rsv(fdt_blob);
 	for (i = 0; i < total; i++) {
 		if (fdt_get_mem_rsv(fdt_blob, i, &addr, &size) != 0)
 			continue;
-		boot_fdt_reserve_region(addr, size, LMB_NOOVERWRITE);
+		boot_fdt_handle_region(addr, size, LMB_NOOVERWRITE, free);
 	}
 
 	/* process reserved-memory */
@@ -131,7 +143,7 @@ void boot_fdt_add_mem_rsv_regions(void *fdt_blob)
 					flags = LMB_NOMAP;
 				addr = res.start;
 				size = res.end - res.start + 1;
-				boot_fdt_reserve_region(addr, size, flags);
+				boot_fdt_handle_region(addr, size, flags, free);
 			}
 
 			subnode = fdt_next_subnode(fdt_blob, subnode);
@@ -139,6 +151,31 @@ void boot_fdt_add_mem_rsv_regions(void *fdt_blob)
 	}
 }
 
+/**
+ * boot_fdt_add_mem_rsv_regions - Add FDT memreserve and reserved-memory
+ *                                sections
+ * @fdt_blob: pointer to fdt blob base address
+ *
+ * Adds reserved-memory and memreserve regions in the dtb to the lmb block.
+ * Adding the memreserve regions prevents u-boot from using them to store the
+ * initrd or the fdt blob.
+ *
+ * This function will attempt to clean the currently active reservations if a
+ * new device tree blob is given. This must be called before gd->fdt_blob is
+ * switched.
+ */
+void boot_fdt_add_mem_rsv_regions(const void *fdt_blob)
+{
+	if (fdt_check_header(fdt_blob) != 0)
+		return;
+
+	/* Remove old regions */
+	if (gd->fdt_blob != fdt_blob)
+		boot_fdt_handle_mem_rsv_regions(gd->fdt_blob, true);
+
+	boot_fdt_handle_mem_rsv_regions(fdt_blob, false);
+}
+
 /**
  * boot_relocate_fdt - relocate flat device tree
  * @of_flat_tree: pointer to a char* variable, will hold fdt start address
include/image.h +1 -1
index 34efac6056d..151619f42bf 100644
@@ -827,7 +827,7 @@ int boot_get_fdt(void *buf, const char *select, uint arch,
 		 struct bootm_headers *images, char **of_flat_tree,
 		 ulong *of_size);
 
-void boot_fdt_add_mem_rsv_regions(void *fdt_blob);
+void boot_fdt_add_mem_rsv_regions(const void *fdt_blob);
 int boot_relocate_fdt(char **of_flat_tree, ulong *of_size);
 
 int boot_ramdisk_high(ulong rd_data, ulong rd_len, ulong *initrd_start,
Signature · 1 line
2.54.0
#2 lore raw

PATCH v8 2/3 test: boot: add a fdt reserved region check

In reply to [email protected]

From: Randolph Sapp <[email protected]>

Add a image_fdt suite and a check for boot_fdt_add_mem_rsv_regions. This will ensure the user is properly informed of any reservation failures. It will also validate that reservations are cleaned up correctly when switching FDTs.

---

v8:
- Rework test_boot_fdt_add_mem_rsv_regions to mitigate lingering state

 test/boot/Makefile          |  1 +
 test/boot/image_fdt.c       | 70 +++++++++++++++++++++++++++++++++++++
 test/cmd_ut.c               |  2 ++
 test/py/tests/test_suite.py |  2 +-
 4 files changed, 74 insertions(+), 1 deletion(-)
 create mode 100644 test/boot/image_fdt.c
test/boot/Makefile +1 -0
index 89538d4f0a6..640af6593cd 100644
@@ -14,6 +14,7 @@ endif
 
 ifdef CONFIG_SANDBOX
 obj-$(CONFIG_$(PHASE_)CMDLINE) += bootm.o
+obj-$(CONFIG_$(PHASE_)OF_LIBFDT) += image_fdt.o
 endif
 obj-$(CONFIG_MEASURED_BOOT) += measurement.o
 
test/boot/image_fdt.c new file +70 -0

70 lines not on this page · raw message

test/cmd_ut.c +2 -0
index 44e5fdfdaa6..363ed4eab30 100644
@@ -61,6 +61,7 @@ SUITE_DECL(fdt);
 SUITE_DECL(fdt_overlay);
 SUITE_DECL(font);
 SUITE_DECL(hush);
+SUITE_DECL(image_fdt);
 SUITE_DECL(lib);
 SUITE_DECL(loadm);
 SUITE_DECL(log);
@@ -88,6 +89,7 @@ static struct suite suites[] = {
 	SUITE(fdt_overlay, "device tree overlays"),
 	SUITE(font, "font command"),
 	SUITE(hush, "hush behaviour"),
+	SUITE(image_fdt, "image fdt parsing"),
 	SUITE(lib, "library functions"),
 	SUITE(loadm, "loadm command parameters and loading memory blob"),
 	SUITE(log, "logging functions"),
test/py/tests/test_suite.py +1 -1
index 7fe9a90dfd3..08285f12a5f 100644
@@ -8,7 +8,7 @@ import re
 EXPECTED_SUITES = [
     'addrmap', 'bdinfo', 'bloblist', 'bootm', 'bootstd',
     'cmd', 'common', 'dm', 'env', 'exit', 'fdt_overlay',
-    'fdt', 'font', 'hush', 'lib',
+    'fdt', 'font', 'hush', 'image_fdt', 'lib',
     'loadm', 'log', 'mbr', 'measurement', 'mem',
     'pci_mps', 'setexpr', 'upl',
     ]
Signature · 1 line
2.54.0
Acked-by on v8 2/3
#3 lore raw

Re: [PATCHv8 2/3] test: boot: add a fdt reserved region check

Quoted 10 lines · On Wed, 13 May 2026 at 23:49, <[email protected]> wrote:

From: Randolph Sapp <[email protected]>

Add a image_fdt suite and a check for boot_fdt_add_mem_rsv_regions. This will ensure the user is properly informed of any reservation failures. It will also validate that reservations are cleaned up correctly when switching FDTs.

---

Quoted 135 lines

v8:
- Rework test_boot_fdt_add_mem_rsv_regions to mitigate lingering state

 test/boot/Makefile          |  1 +
 test/boot/image_fdt.c       | 70 +++++++++++++++++++++++++++++++++++++
 test/cmd_ut.c               |  2 ++
 test/py/tests/test_suite.py |  2 +-
 4 files changed, 74 insertions(+), 1 deletion(-)
 create mode 100644 test/boot/image_fdt.c
test/boot/Makefile +1 -0
index 89538d4f0a6..640af6593cd 100644
@@ -14,6 +14,7 @@ endif
 
 ifdef CONFIG_SANDBOX
 obj-$(CONFIG_$(PHASE_)CMDLINE) += bootm.o
+obj-$(CONFIG_$(PHASE_)OF_LIBFDT) += image_fdt.o
 endif
 obj-$(CONFIG_MEASURED_BOOT) += measurement.o
 
test/boot/image_fdt.c new file +70 -0
new file mode 100644
index 00000000000..778951c1290
@@ -0,0 +1,70 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2026 Texas Instruments Incorporated - https://www.ti.com/
+ */
+
+#include <config.h>
+
+#include <fdt_support.h>
+#include <image.h>
+#include <lmb.h>
+#include <malloc.h>
+
+#include <asm/global_data.h>
+
+#include <test/test.h>
+#include <test/ut.h>
+
+#define IMAGE_FDT_TEST(_name, _flags) UNIT_TEST(_name, _flags, image_fdt)
+
+DECLARE_GLOBAL_DATA_PTR;
+
+static int test_boot_fdt_add_mem_rsv_regions(struct unit_test_state *uts)
+{
+       phys_addr_t start = CFG_SYS_SDRAM_BASE + 0x100000;
+       const void *old_blob = gd->fdt_blob;
+       int ret = CMD_RET_FAILURE;
+       ulong fdt_sz;
+       int nodeoffset;
+       void *new_blob;
+
+       /* Default reservation should exist */
+       ut_asserteq(1, lmb_is_reserved_flags(start, LMB_NOMAP));
+
+       /* Attempting to re-reserve should warn the user */
+       boot_fdt_add_mem_rsv_regions(gd->fdt_blob);
+       ut_assert_nextlinen("ERROR: reserving");
+       ut_assert_console_end();
+
+       /* Loading a new_blob device tree should be allowed */
+       fdt_sz = fdt_totalsize(gd->fdt_blob);
+       new_blob = malloc(fdt_sz);
+       ut_assertok_ptr(new_blob);
+       memcpy(new_blob, gd->fdt_blob, fdt_sz);
+
+       nodeoffset = fdt_path_offset(new_blob, "/reserved-memory");
+       if (nodeoffset < 0)
+               goto free_blob;
+
+       if (fdt_del_node(new_blob, nodeoffset))
+               goto free_blob;
+
+       boot_fdt_add_mem_rsv_regions(new_blob);
+       gd->fdt_blob = new_blob;
+
+       if (ut_assert_console_end())
+               goto switch_fdt;
+
+       /* Reservation should not exist now */
+       if (!lmb_is_reserved_flags(start, LMB_NOMAP))
+               ret = 0;
+
+       /* Cleanup */
+switch_fdt:
+       boot_fdt_add_mem_rsv_regions(old_blob);
+       gd->fdt_blob = old_blob;
+free_blob:
+       free(new_blob);
+       return ret;
+}
+IMAGE_FDT_TEST(test_boot_fdt_add_mem_rsv_regions, UTF_LIVE_TREE | UTF_CONSOLE);
test/cmd_ut.c +2 -0
index 44e5fdfdaa6..363ed4eab30 100644
@@ -61,6 +61,7 @@ SUITE_DECL(fdt);
 SUITE_DECL(fdt_overlay);
 SUITE_DECL(font);
 SUITE_DECL(hush);
+SUITE_DECL(image_fdt);
 SUITE_DECL(lib);
 SUITE_DECL(loadm);
 SUITE_DECL(log);
@@ -88,6 +89,7 @@ static struct suite suites[] = {
        SUITE(fdt_overlay, "device tree overlays"),
        SUITE(font, "font command"),
        SUITE(hush, "hush behaviour"),
+       SUITE(image_fdt, "image fdt parsing"),
        SUITE(lib, "library functions"),
        SUITE(loadm, "loadm command parameters and loading memory blob"),
        SUITE(log, "logging functions"),
test/py/tests/test_suite.py +1 -2
index 7fe9a90dfd3..08285f12a5f 100644
@@ -8,7 +8,7 @@ import re
 EXPECTED_SUITES = [
     'addrmap', 'bdinfo', 'bloblist', 'bootm', 'bootstd',
     'cmd', 'common', 'dm', 'env', 'exit', 'fdt_overlay',
-    'fdt', 'font', 'hush', 'lib',
+    'fdt', 'font', 'hush', 'image_fdt', 'lib',
     'loadm', 'log', 'mbr', 'measurement', 'mem',
     'pci_mps', 'setexpr', 'upl',
     ]
--

2.54.0

Reviewed-by on v8 2/3
#4 lore raw

Re: [PATCHv8 2/3] test: boot: add a fdt reserved region check

In reply to [email protected]

Hi Randolph,

Quoted 7 lines · Randolph Sapp

test: boot: add a fdt reserved region check

Add a image_fdt suite and a check for boot_fdt_add_mem_rsv_regions. This will ensure the user is properly informed of any reservation failures. It will also validate that reservations are cleaned up correctly when switching FDTs.

Quoted 8 lines

test/boot/Makefile | 1 +

 test/boot/image_fdt.c       | 70 +++++++++++++++++++++++++++++++++++++++++++++
 test/cmd_ut.c               |  2 ++
 test/py/tests/test_suite.py |  2 +-
 4 files changed, 74 insertions(+), 1 deletion(-)

test/boot/image_fdt.c line 4

Quoted 6 lines
test/boot/image_fdt.c +4 -0
@@ -0,0 +1,70 @@
+     fdt_sz = fdt_totalsize(gd->fdt_blob);
+     new_blob = malloc(fdt_sz);
+     ut_assertok_ptr(new_blob);
+     memcpy(new_blob, gd->fdt_blob, fdt_sz);

ut_assertok_ptr() only checks IS_ERR(), but malloc() returns NULL on failure. IS_ERR(NULL) is false, so a failed allocation sails through and crashes in the memcpy() below. Please use ut_assertnonnull(new_blob) here.

test/boot/image_fdt.c line 5

Quoted 7 lines
test/boot/image_fdt.c +5 -0
@@ -0,0 +1,70 @@
+     boot_fdt_add_mem_rsv_regions(new_blob);
+     gd->fdt_blob = new_blob;
+
+     if (ut_assert_console_end())
+             goto switch_fdt;

This doesn't do what it looks like. ut_assert_console_end() expands to a block that does 'return CMD_RET_FAILURE' on failure, so the if condition is always 0 and the goto is dead code. If the assertion fails, the function returns immediately, leaving gd->fdt_blob pointing at new_blob and leaking it.

To defer cleanup, either restructure so gd->fdt_blob isn't changed until after the check, or use ut_check_console_end() and handle the result yourself. The same applies to the ut_assertok_ptr()/ut_assertnonnull() above - once gd->fdt_blob is mutated, any macro that early-returns skips the cleanup labels. We have a similar problem in the bootstd tests...

Regards,
Simon

#5 lore raw

PATCH v8 3/3 memory: reserve from start_addr_sp to initial_relocaddr

In reply to [email protected]

From: Randolph Sapp <[email protected]>

Add a new global data struct member called initial_relocaddr. This stores the original value of relocaddr, directly from setup_dest_addr. This is specifically to avoid any adjustments made by other init functions.

Reserve the memory from gd->start_addr_sp - CONFIG_STACK_SIZE to gd->initial_relocaddr instead of gd->ram_top. This allows platform specific relocation addresses to work without unnecessarily painting over a large range.

---

v7:
- Reinstate bank hopping logic for U-Boot reserved region
- Update description for gd->initial_relocaddr, make it clear this is also an exclusive value
- Leave the PRAM region out of the reservation. Previous commit messages indicate that this is intended.
v8:
- Adjust initial_relocaddr doc string

 common/board_f.c                  | 9 ++++++++-
 include/asm-generic/global_data.h | 9 +++++++++
 lib/efi_loader/efi_memory.c       | 2 +-
 lib/lmb.c                         | 7 ++++---
 4 files changed, 22 insertions(+), 5 deletions(-)
common/board_f.c +8 -1
index ce87c619e68..aeb53b4c274 100644
@@ -330,6 +330,8 @@ __weak int arch_setup_dest_addr(void)
 
 static int setup_dest_addr(void)
 {
+	int ret;
+
 	debug("Monitor len: %08x\n", gd->mon_len);
 	/*
 	 * Ram is setup, size stored in gd !!
@@ -356,7 +358,12 @@ static int setup_dest_addr(void)
 	gd->relocaddr = gd->ram_top;
 	debug("Ram top: %08llX\n", (unsigned long long)gd->ram_top);
 
-	return arch_setup_dest_addr();
+	ret = arch_setup_dest_addr();
+	if (ret)
+		return ret;
+
+	gd->initial_relocaddr = gd->relocaddr;
+	return 0;
 }
 
 #ifdef CFG_PRAM
include/asm-generic/global_data.h +9 -0
index 745d2c3a966..8d1d49b1133 100644
@@ -107,6 +107,15 @@ struct global_data {
 	 * GDB using the 'add-symbol-file u-boot <relocaddr>' command.
 	 */
 	unsigned long relocaddr;
+	/**
+	 * @initial_relocaddr: top address of U-Boot in RAM
+	 *
+	 * This should be the value of relocaddr after setup_dest_addr() and
+	 * before reserve_pram() or any other allocations or reservations shift
+	 * it. This address will, depending on the platform, be equivalent to
+	 * ram_top and should also be considered an exclusive address.
+	 */
+	unsigned long initial_relocaddr;
 	/**
 	 * @irq_sp: IRQ stack pointer
 	 */
lib/efi_loader/efi_memory.c +1 -1
index 046a2bb4641..9d7eda9422f 100644
@@ -869,7 +869,7 @@ static void add_u_boot_and_runtime(void)
 	/* Add U-Boot */
 	uboot_start = ((uintptr_t)map_sysmem(gd->start_addr_sp, 0) -
 		       uboot_stack_size) & ~EFI_PAGE_MASK;
-	uboot_pages = ((uintptr_t)map_sysmem(gd->ram_top - 1, 0) -
+	uboot_pages = ((uintptr_t)map_sysmem(gd->initial_relocaddr - 1, 0) -
 		       uboot_start + EFI_PAGE_MASK) >> EFI_PAGE_SHIFT;
 	efi_update_memory_map(uboot_start, uboot_pages, EFI_BOOT_SERVICES_CODE,
 			      false, false);
lib/lmb.c +4 -3
index 8f12c6ad8e5..27c8565e590 100644
@@ -540,13 +540,14 @@ static void lmb_reserve_uboot_region(void)
 	ulong pram = 0;
 
 	rsv_start = gd->start_addr_sp - CONFIG_STACK_SIZE;
-	end = gd->ram_top;
+	end = gd->initial_relocaddr;
 
 	/*
 	 * Reserve memory from aligned address below the bottom of U-Boot stack
-	 * until end of RAM area to prevent LMB from overwriting that memory.
+	 * until the original relocation address to prevent LMB from
+	 * overwriting that memory.
 	 */
-	debug("## Current stack ends at 0x%08lx ", (ulong)rsv_start);
+	debug("## Current stack ends at 0x%08lx\n", (ulong)rsv_start);
 
 #ifdef CFG_PRAM
 	pram = env_get_ulong("pram", 10, CFG_PRAM);
Signature · 1 line
2.54.0
Reviewed-by on v8 3/3
#6 lore raw

Re: [PATCHv8 3/3] memory: reserve from start_addr_sp to initial_relocaddr

Quoted 8 lines · On Wed, 13 May 2026 at 23:49, <[email protected]> wrote:

From: Randolph Sapp <[email protected]>

Add a new global data struct member called initial_relocaddr. This stores the original value of relocaddr, directly from setup_dest_addr. This is specifically to avoid any adjustments made by other init functions.

Quoted 8 lines

Reserve the memory from gd->start_addr_sp - CONFIG_STACK_SIZE to gd->initial_relocaddr instead of gd->ram_top. This allows platform specific relocation addresses to work without unnecessarily painting over a large range.

---

Quoted 101 lines

v7:
- Reinstate bank hopping logic for U-Boot reserved region
- Update description for gd->initial_relocaddr, make it clear this is also an exclusive value
- Leave the PRAM region out of the reservation. Previous commit messages indicate that this is intended.
v8:
- Adjust initial_relocaddr doc string

 common/board_f.c                  | 9 ++++++++-
 include/asm-generic/global_data.h | 9 +++++++++
 lib/efi_loader/efi_memory.c       | 2 +-
 lib/lmb.c                         | 7 ++++---
 4 files changed, 22 insertions(+), 5 deletions(-)
common/board_f.c +8 -1
index ce87c619e68..aeb53b4c274 100644
@@ -330,6 +330,8 @@ __weak int arch_setup_dest_addr(void)
 
 static int setup_dest_addr(void)
 {
+       int ret;
+
        debug("Monitor len: %08x\n", gd->mon_len);
        /*
         * Ram is setup, size stored in gd !!
@@ -356,7 +358,12 @@ static int setup_dest_addr(void)
        gd->relocaddr = gd->ram_top;
        debug("Ram top: %08llX\n", (unsigned long long)gd->ram_top);
 
-       return arch_setup_dest_addr();
+       ret = arch_setup_dest_addr();
+       if (ret)
+               return ret;
+
+       gd->initial_relocaddr = gd->relocaddr;
+       return 0;
 }
 
 #ifdef CFG_PRAM
include/asm-generic/global_data.h +9 -0
index 745d2c3a966..8d1d49b1133 100644
@@ -107,6 +107,15 @@ struct global_data {
         * GDB using the 'add-symbol-file u-boot <relocaddr>' command.
         */
        unsigned long relocaddr;
+       /**
+        * @initial_relocaddr: top address of U-Boot in RAM
+        *
+        * This should be the value of relocaddr after setup_dest_addr() and
+        * before reserve_pram() or any other allocations or reservations shift
+        * it. This address will, depending on the platform, be equivalent to
+        * ram_top and should also be considered an exclusive address.
+        */
+       unsigned long initial_relocaddr;
        /**
         * @irq_sp: IRQ stack pointer
         */
lib/efi_loader/efi_memory.c +1 -1
index 046a2bb4641..9d7eda9422f 100644
@@ -869,7 +869,7 @@ static void add_u_boot_and_runtime(void)
        /* Add U-Boot */
        uboot_start = ((uintptr_t)map_sysmem(gd->start_addr_sp, 0) -
                       uboot_stack_size) & ~EFI_PAGE_MASK;
-       uboot_pages = ((uintptr_t)map_sysmem(gd->ram_top - 1, 0) -
+       uboot_pages = ((uintptr_t)map_sysmem(gd->initial_relocaddr - 1, 0) -
                       uboot_start + EFI_PAGE_MASK) >> EFI_PAGE_SHIFT;
        efi_update_memory_map(uboot_start, uboot_pages, EFI_BOOT_SERVICES_CODE,
                              false, false);
lib/lmb.c +4 -4
index 8f12c6ad8e5..27c8565e590 100644
@@ -540,13 +540,14 @@ static void lmb_reserve_uboot_region(void)
        ulong pram = 0;
 
        rsv_start = gd->start_addr_sp - CONFIG_STACK_SIZE;
-       end = gd->ram_top;
+       end = gd->initial_relocaddr;
 
        /*
         * Reserve memory from aligned address below the bottom of U-Boot stack
-        * until end of RAM area to prevent LMB from overwriting that memory.
+        * until the original relocation address to prevent LMB from
+        * overwriting that memory.
         */
-       debug("## Current stack ends at 0x%08lx ", (ulong)rsv_start);
+       debug("## Current stack ends at 0x%08lx\n", (ulong)rsv_start);
 
 #ifdef CFG_PRAM
        pram = env_get_ulong("pram", 10, CFG_PRAM);
--

2.54.0

#7 lore raw
In reply to [email protected]
Quoted 8 lines · On Wed, May 13, 2026 at 03:49:40PM -0500, [email protected] wrote:

From: Randolph Sapp <[email protected]>

Nitpicks and fixes from the discovery thread on adding PocketBeagle2 support [1]. This does a lot of general setup required for the device, but these modifications themselves aren't device specific. For those specifically interested in PocketBeagle2 support and don't care about these details, my development branch is public [2].

Quoted 7 lines

That first patch may provoke some opinions, but honestly if that warning was still present I wouldn't have spent a week poking holes in both the EFI and LMB allocations systems. Please let me know if there is a specific usecase that it breaks though.

[1] https://lore.kernel.org/all/[email protected]/
[2] https://github.com/StaticRocket/u-boot/tree/feature/pocketbeagle2

This leads to test failures now (that either need to be fixed or updated):
https://source.denx.de/u-boot/u-boot/-/jobs/1464028 and please see
https://docs.u-boot.org/en/latest/develop/ci_testing.html as well for starting CI, however you might want to run the tests locally first. Thanks.

Signature · 1 line
Tom
#8 lore raw
Quoted 18 lines · Tom Rini
Quoted 8 lines · On Wed, May 13, 2026 at 03:49:40PM -0500, [email protected] wrote:

From: Randolph Sapp <[email protected]>

Nitpicks and fixes from the discovery thread on adding PocketBeagle2 support [1]. This does a lot of general setup required for the device, but these modifications themselves aren't device specific. For those specifically interested in PocketBeagle2 support and don't care about these details, my development branch is public [2].

Quoted 7 lines

That first patch may provoke some opinions, but honestly if that warning was still present I wouldn't have spent a week poking holes in both the EFI and LMB allocations systems. Please let me know if there is a specific usecase that it breaks though.

[1] https://lore.kernel.org/all/[email protected]/
[2] https://github.com/StaticRocket/u-boot/tree/feature/pocketbeagle2

Quoted 6 lines

This leads to test failures now (that either need to be fixed or updated):
https://source.denx.de/u-boot/u-boot/-/jobs/1464028 and please see
https://docs.u-boot.org/en/latest/develop/ci_testing.html as well for starting CI, however you might want to run the tests locally first. Thanks.

So, that's the fun part. I ran qcheck with this. I went tracing through that make target and even started manually issuing "./test/py/test.py --bd sandbox_flattree --build -k ut_image_fdt_test_boot_fdt_add_mem_rsv_regions -s" (and again with sandbox, sandbox64, etc.) to see exactly why that's failing in CI. I cannot reproduce the issue reported. It passes and reports all the messages I'm expecting on my end.

The CI pipeline must not be loading the test.dtb. No idea how that's occurring, or how the other UT_DM tests are passing. I don't see any particularly weird parameters in the test logs.

It's not like this test is dependent on the same things UTF_DM tagged tests are, but I expected to be able to reuse that device tree without having to specify that flag. That's the behavior I have locally at least. Adding an explicit check for CONFIG_UT_DM didn't seem to help at all.

Any advice here?

#9 lore raw
Quoted 37 lines · Randolph Sapp
Quoted 18 lines · Tom Rini
Quoted 8 lines · On Wed, May 13, 2026 at 03:49:40PM -0500, [email protected] wrote:

From: Randolph Sapp <[email protected]>

Nitpicks and fixes from the discovery thread on adding PocketBeagle2 support [1]. This does a lot of general setup required for the device, but these modifications themselves aren't device specific. For those specifically interested in PocketBeagle2 support and don't care about these details, my development branch is public [2].

Quoted 7 lines

That first patch may provoke some opinions, but honestly if that warning was still present I wouldn't have spent a week poking holes in both the EFI and LMB allocations systems. Please let me know if there is a specific usecase that it breaks though.

[1] https://lore.kernel.org/all/[email protected]/
[2] https://github.com/StaticRocket/u-boot/tree/feature/pocketbeagle2

Quoted 6 lines

This leads to test failures now (that either need to be fixed or updated):
https://source.denx.de/u-boot/u-boot/-/jobs/1464028 and please see
https://docs.u-boot.org/en/latest/develop/ci_testing.html as well for starting CI, however you might want to run the tests locally first. Thanks.

So, that's the fun part. I ran qcheck with this. I went tracing through that make target and even started manually issuing "./test/py/test.py --bd sandbox_flattree --build -k ut_image_fdt_test_boot_fdt_add_mem_rsv_regions -s" (and again with sandbox, sandbox64, etc.) to see exactly why that's failing in CI. I cannot reproduce the issue reported. It passes and reports all the messages I'm expecting on my end.

The CI pipeline must not be loading the test.dtb. No idea how that's occurring, or how the other UT_DM tests are passing. I don't see any particularly weird parameters in the test logs.

Quoted 6 lines

It's not like this test is dependent on the same things UTF_DM tagged tests are, but I expected to be able to reuse that device tree without having to specify that flag. That's the behavior I have locally at least. Adding an explicit check for CONFIG_UT_DM didn't seem to help at all.

Any advice here?

Have you tried running it the way it's shown in the .azure-pipelines.yml or .gitlab-ci.yml files? And I wouldn't limit things to just the single test, it's possible there's some unexpected thing to be cleaned up from another test that's contaminating things.

Signature · 1 line
Tom
#10 lore raw
Quoted 43 lines · Tom Rini
Quoted 42 lines · Randolph Sapp
Quoted 18 lines · Tom Rini
Quoted 8 lines · On Wed, May 13, 2026 at 03:49:40PM -0500, [email protected] wrote:

From: Randolph Sapp <[email protected]>

Nitpicks and fixes from the discovery thread on adding PocketBeagle2 support [1]. This does a lot of general setup required for the device, but these modifications themselves aren't device specific. For those specifically interested in PocketBeagle2 support and don't care about these details, my development branch is public [2].

Quoted 7 lines

That first patch may provoke some opinions, but honestly if that warning was still present I wouldn't have spent a week poking holes in both the EFI and LMB allocations systems. Please let me know if there is a specific usecase that it breaks though.

[1] https://lore.kernel.org/all/[email protected]/
[2] https://github.com/StaticRocket/u-boot/tree/feature/pocketbeagle2

Quoted 6 lines

This leads to test failures now (that either need to be fixed or updated):
https://source.denx.de/u-boot/u-boot/-/jobs/1464028 and please see
https://docs.u-boot.org/en/latest/develop/ci_testing.html as well for starting CI, however you might want to run the tests locally first. Thanks.

So, that's the fun part. I ran qcheck with this. I went tracing through that make target and even started manually issuing "./test/py/test.py --bd sandbox_flattree --build -k ut_image_fdt_test_boot_fdt_add_mem_rsv_regions -s" (and again with sandbox, sandbox64, etc.) to see exactly why that's failing in CI. I cannot reproduce the issue reported. It passes and reports all the messages I'm expecting on my end.

The CI pipeline must not be loading the test.dtb. No idea how that's occurring, or how the other UT_DM tests are passing. I don't see any particularly weird parameters in the test logs.

It's not like this test is dependent on the same things UTF_DM tagged tests are, but I expected to be able to reuse that device tree without having to specify that flag. That's the behavior I have locally at least. Adding an explicit check for CONFIG_UT_DM didn't seem to help at all.

Quoted 6 lines
Quoted 1 line

Any advice here?

Have you tried running it the way it's shown in the .azure-pipelines.yml or .gitlab-ci.yml files? And I wouldn't limit things to just the single test, it's possible there's some unexpected thing to be cleaned up from another test that's contaminating things.

Hell. Had to bisect all of the ut tests just to find out ut_bootstd_bootflow_cmd_boot is changing out the device tree before my test.

I didn't want to make my test dynamic because there's a chance any reservations I try to make could overlap with a device or some already allocated range. I need some known working allocation that's in a valid DT to be safe.

The bootflow_cmd_boot test will change the state of the running sandbox in a way that I assume can not be easily reverted. Please inform me if this assumption is untrue.

I can add a wrapper fixture in pytest to make sure the sandbox is restarted specifically after that test, but I'm not sure if that's appropriate.

---

As an aside, test_ut_dm_init_bootstd should absolutely be a fixture instead of a test. May need to fix that later.

#11 lore raw
Quoted 62 lines · Randolph Sapp
Quoted 43 lines · Tom Rini
Quoted 42 lines · Randolph Sapp
Quoted 18 lines · Tom Rini
Quoted 8 lines · On Wed, May 13, 2026 at 03:49:40PM -0500, [email protected] wrote:

From: Randolph Sapp <[email protected]>

Nitpicks and fixes from the discovery thread on adding PocketBeagle2 support [1]. This does a lot of general setup required for the device, but these modifications themselves aren't device specific. For those specifically interested in PocketBeagle2 support and don't care about these details, my development branch is public [2].

Quoted 7 lines

That first patch may provoke some opinions, but honestly if that warning was still present I wouldn't have spent a week poking holes in both the EFI and LMB allocations systems. Please let me know if there is a specific usecase that it breaks though.

[1] https://lore.kernel.org/all/[email protected]/
[2] https://github.com/StaticRocket/u-boot/tree/feature/pocketbeagle2

Quoted 6 lines

This leads to test failures now (that either need to be fixed or updated):
https://source.denx.de/u-boot/u-boot/-/jobs/1464028 and please see
https://docs.u-boot.org/en/latest/develop/ci_testing.html as well for starting CI, however you might want to run the tests locally first. Thanks.

So, that's the fun part. I ran qcheck with this. I went tracing through that make target and even started manually issuing "./test/py/test.py --bd sandbox_flattree --build -k ut_image_fdt_test_boot_fdt_add_mem_rsv_regions -s" (and again with sandbox, sandbox64, etc.) to see exactly why that's failing in CI. I cannot reproduce the issue reported. It passes and reports all the messages I'm expecting on my end.

The CI pipeline must not be loading the test.dtb. No idea how that's occurring, or how the other UT_DM tests are passing. I don't see any particularly weird parameters in the test logs.

It's not like this test is dependent on the same things UTF_DM tagged tests are, but I expected to be able to reuse that device tree without having to specify that flag. That's the behavior I have locally at least. Adding an explicit check for CONFIG_UT_DM didn't seem to help at all.

Quoted 6 lines
Quoted 1 line

Any advice here?

Have you tried running it the way it's shown in the .azure-pipelines.yml or .gitlab-ci.yml files? And I wouldn't limit things to just the single test, it's possible there's some unexpected thing to be cleaned up from another test that's contaminating things.

Hell. Had to bisect all of the ut tests just to find out ut_bootstd_bootflow_cmd_boot is changing out the device tree before my test.

I didn't want to make my test dynamic because there's a chance any reservations I try to make could overlap with a device or some already allocated range. I need some known working allocation that's in a valid DT to be safe.

The bootflow_cmd_boot test will change the state of the running sandbox in a way that I assume can not be easily reverted. Please inform me if this assumption is untrue.

Quoted 7 lines

I can add a wrapper fixture in pytest to make sure the sandbox is restarted specifically after that test, but I'm not sure if that's appropriate.

---

As an aside, test_ut_dm_init_bootstd should absolutely be a fixture instead of a test. May need to fix that later.

There's other tests which force a reboot, and so it wouldn't surprise me if one got missed. And then yes, our pytests themselves would benefit from someone with a bit better understanding of pytest spending time on it, or just someone spending time on them and not a dozen other things at the same time as well. Thanks!

Signature · 1 line
Tom