Re: [PATCH v8 01/17] memblock: add MEMBLOCK_RSRV_KERN flag
From: Pratyush Yadav <pratyush@kernel.org>
Date: 2025-10-13 16:40:16
Also in:
kexec, linux-arm-kernel, linux-doc, linux-mm, lkml
On Mon, Oct 13 2025, Pratyush Yadav wrote:
On Fri, Oct 10 2025, Breno Leitao wrote:quoted
Hello Chanyuan, Mike, On Fri, May 09, 2025 at 12:46:19AM -0700, Changyuan Lyu wrote:quoted
--- a/mm/memblock.c +++ b/mm/memblock.c@@ -492,7 +492,7 @@ static int __init_memblock memblock_double_array(struct memblock_type *type, * needn't do it */ if (!use_slab) - BUG_ON(memblock_reserve(addr, new_alloc_size)); + BUG_ON(memblock_reserve_kern(addr, new_alloc_size)); /* Update slab flag */ *in_slab = use_slab;@@ -642,7 +642,7 @@ static int __init_memblock memblock_add_range(struct memblock_type *type, #ifdef CONFIG_NUMA WARN_ON(nid != memblock_get_region_node(rgn)); #endif - WARN_ON(flags != rgn->flags); + WARN_ON(flags != MEMBLOCK_NONE && flags != rgn->flags);I am hitting some sporadic warnings at early boot on a production kernel (6.16). Unfortunately this issue is not easily reproduce for me to test on upstream. 09:14:44 BIOS-provided physical RAM map: 09:14:44 BIOS-e820: [mem 0x0000000000000000-0x000000000009ffff] usable 09:14:44 BIOS-e820: [mem 0x00000000000a0000-0x00000000000fffff] reserved 09:14:44 BIOS-e820: [mem 0x0000000000100000-0x0000000064cb7fff] usable 09:14:44 BIOS-e820: [mem 0x0000000064cb8000-0x0000000064dc3fff] reserved 09:14:44 BIOS-e820: [mem 0x0000000064dc4000-0x0000000065b13fff] usable 09:14:44 BIOS-e820: [mem 0x0000000065b14000-0x0000000065b61fff] reserved 09:14:44 BIOS-e820: [mem 0x0000000065b62000-0x0000000065ed0fff] usable 09:14:44 BIOS-e820: [mem 0x0000000065ed1000-0x0000000065f2bfff] reserved 09:14:44 BIOS-e820: [mem 0x0000000065f2c000-0x0000000066621fff] usable 09:14:44 BIOS-e820: [mem 0x0000000066622000-0x0000000066630fff] reserved 09:14:44 BIOS-e820: [mem 0x0000000066631000-0x0000000068107fff] usable 09:14:44 BIOS-e820: [mem 0x0000000068108000-0x000000006819dfff] ACPI data 09:14:44 BIOS-e820: [mem 0x000000006819e000-0x000000006a48cfff] usable 09:14:44 BIOS-e820: [mem 0x000000006a48d000-0x000000006c58cfff] reserved 09:14:44 BIOS-e820: [mem 0x000000006c58d000-0x000000006c5dcfff] ACPI data 09:14:44 BIOS-e820: [mem 0x000000006c5dd000-0x000000006cfdcfff] ACPI NVS 09:14:44 BIOS-e820: [mem 0x000000006cfdd000-0x000000006e9fcfff] reserved 09:14:44 BIOS-e820: [mem 0x000000006e9fd000-0x000000006fffffff] usable 09:14:44 BIOS-e820: [mem 0x0000000070000000-0x000000008fffffff] reserved 09:14:44 BIOS-e820: [mem 0x00000000fd000000-0x00000000fe7fffff] reserved 09:14:44 BIOS-e820: [mem 0x00000000fed20000-0x00000000fed44fff] reserved 09:14:44 BIOS-e820: [mem 0x00000000ff000000-0x00000000ffffffff] reserved 09:14:44 BIOS-e820: [mem 0x0000000100000000-0x000000107fff655f] usable 09:14:44 BIOS-e820: [mem 0x000000107fff6560-0x000000107fff656f] type 128 09:14:44 BIOS-e820: [mem 0x000000107fff6570-0x000000107fffffff] usable 09:14:44 random: crng init done 09:14:44 ------------[ cut here ]------------ 09:14:44 WARNING: CPU: 0 PID: 0 at mm/memblock.c:668 memblock_add_range (mm/memblock.c:668) 09:14:44 Modules linked in: 09:14:44 Tainted: [S]=CPU_OUT_OF_SPEC 09:14:44 RIP: 0010:memblock_add_range (mm/memblock.c:668) 09:14:44 Code: 28 80 3c 01 00 0f 84 04 fd ff ff 4c 89 ef e8 6c 77 09 00 e9 f7 fc ff ff 0f 0b 83 7c 24 1c 00 0f 85 9c fd ff ff e9 c5 fd ff ff <0f> 0b e9 be fd ff ff 44 89 f1 80 e1 07 80 c1 03 38 c1 0f 8c 6b fd All code ======== 0: 28 80 3c 01 00 0f sub %al,0xf00013c(%rax) 6: 84 04 fd ff ff 4c 89 test %al,-0x76b30001(,%rdi,8) d: ef out %eax,(%dx) e: e8 6c 77 09 00 call 0x9777f 13: e9 f7 fc ff ff jmp 0xfffffffffffffd0f 18: 0f 0b ud2 1a: 83 7c 24 1c 00 cmpl $0x0,0x1c(%rsp) 1f: 0f 85 9c fd ff ff jne 0xfffffffffffffdc1 25: e9 c5 fd ff ff jmp 0xfffffffffffffdef 2a:* 0f 0b ud2 <-- trapping instruction 2c: e9 be fd ff ff jmp 0xfffffffffffffdef 31: 44 89 f1 mov %r14d,%ecx 34: 80 e1 07 and $0x7,%cl 37: 80 c1 03 add $0x3,%cl 3a: 38 c1 cmp %al,%cl 3c: 0f .byte 0xf 3d: 8c 6b fd mov %gs,-0x3(%rbx) Code starting with the faulting instruction =========================================== 0: 0f 0b ud2 2: e9 be fd ff ff jmp 0xfffffffffffffdc5 7: 44 89 f1 mov %r14d,%ecx a: 80 e1 07 and $0x7,%cl d: 80 c1 03 add $0x3,%cl 10: 38 c1 cmp %al,%cl 12: 0f .byte 0xf 13: 8c 6b fd mov %gs,-0x3(%rbx) 09:14:44 RSP: 0000:ffffffff85e07d48 EFLAGS: 00010083 ORIG_RAX: 0000000000000000 09:14:44 RAX: 0000000000000020 RBX: 0000000000001c00 RCX: dffffc0000000000 09:14:44 RDX: 000000000009f000 RSI: 000000000009d000 RDI: ffffffff8685ebf8 09:14:44 RBP: 0000000000000002 R08: 0000000000000020 R09: 0000000000000000 09:14:44 R10: ffffffffff200570 R11: fffffbffffe400b2 R12: 000000000009d000 09:14:44 R13: 0000000000100000 R14: ffffffff8edf5ce4 R15: ffffffff8edf5ce0 09:14:44 FS: 0000000000000000(0000) GS:0000000000000000(0000) knlGS:0000000000000000 09:14:44 CR2: ffff888059e2dff8 CR3: 000000005bc1d000 CR4: 00000000000000b0 09:14:44 Call Trace: 09:14:44 <TASK> 09:14:44 ? __memblock_reserve (mm/memblock.c:936) 09:14:44 ? add_early_ima_buffer (arch/x86/kernel/setup.c:413) 09:14:44 ? parse_setup_data (arch/x86/kernel/setup.c:500) 09:14:44 ? setup_arch (arch/x86/kernel/setup.c:245 arch/x86/kernel/setup.c:958) 09:14:44 ? start_kernel (init/main.c:922) 09:14:44 ? x86_64_start_reservations (arch/x86/kernel/ebda.c:57) 09:14:44 ? x86_64_start_kernel (arch/x86/kernel/head64.c:231) 09:14:44 ? common_startup_64 (arch/x86/kernel/head_64.S:419) 09:14:44 </TASK> .... Memory: 49640988K/66772816K available (54946K kernel code, 19058K rwdata, 22636K rodata, 2940K init, 120968K bss, 10650188K reserved, 6291456K cma-reserved) So, there is a memory override, and I am curious about it. Do you think itYeah, it seems IMA is reserving a region that overlaps a region reserved by something else that doesn't use memblock_reserve_kern().quoted
would be useful to expand this warning to dump more information about the issue? (only compiled tested) if (flags != MEMBLOCK_NONE && flags != rgn->flags) { pr_warn("memblock: Flag mismatch at region [%pa-%pa]\n", &rgn->base, &rend); pr_warn(" Existing region flags: %#x\n", rgn->flags); pr_warn(" New range flags: %#x\n", flags); pr_warn(" New range: [%pa-%pa]\n", &base, &end); WARN_ON_ONCE(1); }I suppose this would be useful. I think enabling memblock debug prints would also be helpful (using the "memblock=debug" commandline parameter) if it doesn't impact your production environment too much.
Actually, I think "memblock=debug" is going to be the more useful thing since it would also show what function allocated the overlapping range and the flags it was allocated with. On my qemu VM with KVM, this results in around 70 prints from memblock. So it adds a bit of extra prints but nothing that should be too disrupting I think. Plus, only at boot so the worst thing you get is slightly slower boot times. -- Regards, Pratyush Yadav