[Crash-utility] Re: [PATCH] bugfix command "help -r" segv fault

Hi, Guanyou Thank you for the fix. On Mon, Nov 4, 2024 at 4:13 PM <devel-request(a)lists.crash-utility.osci.io&gt; wrote: > Date: Fri, 1 Nov 2024 18:01:27 +0800 > From: Guanyou Chen <chenguanyou9338(a)gmail.com&gt; > Subject: [Crash-utility] [PATCH] bugfix command "help -r" segv fault > To: Lianbo <lijiang(a)redhat.com&gt;, Tao Liu <ltao(a)redhat.com&gt;, > devel(a)lists.crash-utility.osci.io > Message-ID: > <CAHS3RMU3nuiqW4z= > Qo9RoufADrUxcaLhyjnxwMCuGODB_+37yQ(a)mail.gmail.com&gt; > Content-Type: multipart/mixed; boundary="00000000000065fc530625d705b8" > > --00000000000065fc530625d705b8 > Content-Type: multipart/alternative; > boundary="00000000000065fc530625d705b6" > > --00000000000065fc530625d705b6 > Content-Type: text/plain; charset="UTF-8" > > Hi Lianbo, Tao > > When the ELF Note does not contain CPU registers, > attempting to retrieve online CPU registers will cause a crash. > > After: > CPU 6: > help: registers not collected for cpu 6 > ... > > Signed-off-by: Guanyou.Chen <chenguanyou(a)xiaomi.com&gt; > --- > netdump.c | 16 ++++++++++++++++ > 1 file changed, 16 insertions(+) > > diff --git a/netdump.c b/netdump.c > index 8ea5159..435793b 100644 > --- a/netdump.c > +++ b/netdump.c > @@ -2780,6 +2780,10 @@ display_regs_from_elf_notes(int cpu, FILE *ofp) > I copied the code block here: display_regs_from_elf_notes(int cpu, FILE *ofp) { Elf32_Nhdr *note32; Elf64_Nhdr *note64; size_t len; char *user_regs; int c, skipped_count; /* * Kdump NT_PRSTATUS notes are only related to online cpus, * so offline cpus should be skipped. */ if (pc->flags2 & QEMU_MEM_DUMP_ELF) skipped_count = 0; else { for (c = skipped_count = 0; c < cpu; c++) { if (check_offline_cpu(c)) skipped_count++; } } if ((cpu - skipped_count) >= nd->num_prstatus_notes && !machine_type("MIPS")) { error(INFO, "registers not collected for cpu %d\n", cpu); return; } ... Could you please point out why the above check does not work? BTW: I'm not sure if it can work for you, can you help to try this? Just a guess. if (((cpu < 0 ) || (!dd->nt_prstatus_percpu[cpu]) || (cpu - skipped_count) >= nd->num_prstatus_notes) && !machine_type("MIPS")) { error(INFO, "registers not collected for cpu %d\n", cpu); return; } Thanks Lianbo nd->nt_prstatus_percpu[cpu]; > else > note64 = (Elf64_Nhdr *)nd->nt_prstatus; > + if (!note64) { > + error(INFO, "registers not collected for cpu %d\n", cpu); > + return; > + } > len = sizeof(Elf64_Nhdr); > len = roundup(len + note64->n_namesz, 4); > len = roundup(len + note64->n_descsz, 4); > @@ -2820,6 +2824,10 @@ display_regs_from_elf_notes(int cpu, FILE *ofp) > nd->nt_prstatus_percpu[cpu]; > else > note32 = (Elf32_Nhdr *)nd->nt_prstatus; > + if (!note32) { > + error(INFO, "registers not collected for cpu %d\n", cpu); > + return; > + } > len = sizeof(Elf32_Nhdr); > len = roundup(len + note32->n_namesz, 4); > len = roundup(len + note32->n_descsz, 4); > @@ -2857,6 +2865,10 @@ display_regs_from_elf_notes(int cpu, FILE *ofp) > else > note64 = (Elf64_Nhdr *)nd->nt_prstatus; > > + if (!note64) { > + error(INFO, "registers not collected for cpu %d\n", cpu); > + return; > + } > prs = (struct ppc64_elf_prstatus *) > ((char *)note64 + sizeof(Elf64_Nhdr) + note64->n_namesz); > prs = (struct ppc64_elf_prstatus *)roundup((ulong)prs, 4); > @@ -2903,6 +2915,10 @@ display_regs_from_elf_notes(int cpu, FILE *ofp) > nd->nt_prstatus_percpu[cpu]; > else > note64 = (Elf64_Nhdr *)nd->nt_prstatus; > + if (!note64) { > + error(INFO, "registers not collected for cpu %d\n", cpu); > + return; > + } > len = sizeof(Elf64_Nhdr); > len = roundup(len + note64->n_namesz, 4); > len = roundup(len + note64->n_descsz, 4); > -- > 2.34.1 > > Guanyou. > Thanks >