Summary: | sys-kernel/hardened-sources-3.4.7 panic on UP->SMP transition when putting CPU core online - BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 IP: [<ffffffff8105a31d>] __setup_vector_irq+0xed/0x130 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Jaak Ristioja <jaak> |
Component: | [OLD] Core system | Assignee: | Anthony Basile <blueness> |
Status: | RESOLVED OBSOLETE | ||
Severity: | normal | CC: | hardened, kernel, pageexec |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: | kernel-config-3.4.7.gz |
Description
Jaak Ristioja
2012-08-12 08:15:51 UTC
Created attachment 321078 [details]
kernel-config-3.4.7.gz
The /proc/config.gz file.
3.4.7-hardened-arm I assume that name doesn't reflect the actual processor architecture in use? (In reply to comment #2) > 3.4.7-hardened-arm > > I assume that name doesn't reflect the actual processor architecture in use? Its amd64 from the panic. @Jaak. A few things before passing this to hardened upstream. Can you upload 1) Your bzImage 2) Your vmlinux. It should be in the root of /usr/src/linux 3) Your System.map If you can, check to see if vanilla 3.4.7 gives the same panic. is this still a problem with 3.7? Please reopen if this is still a problem (In reply to comment #2) > I assume that name doesn't reflect the actual processor architecture in use? Yes, I'm on amd64 (Intel Core2 Duo P8700). "Arm" is Estonian for "grace". I'm just grateful to God for providing me the hardware. :) (In reply to comment #3) > @Jaak. A few things before passing this to hardened upstream. Can you > upload > > 1) Your bzImage > > 2) Your vmlinux. It should be in the root of /usr/src/linux > > 3) Your System.map > > > If you can, check to see if vanilla 3.4.7 gives the same panic. Hmm... For some reason, I didn't receive any notification emails about any comments to this bug. The first and only notification email was for comment #5. I even checked my spam folders and email server logs, nothing else for this bug ever hit my mailbox. So... since a long time has passed since, I don't have these files any more. Sorry. (In reply to comment #4) > is this still a problem with 3.7? When I reported the bug was the only time I have experienced it. I'm currently unable to produce this with hardened-sources-3.8.6. (In reply to comment #6) > (In reply to comment #2) > > I assume that name doesn't reflect the actual processor architecture in use? > > Yes, I'm on amd64 (Intel Core2 Duo P8700). "Arm" is Estonian for "grace". > I'm just grateful to God for providing me the hardware. :) Haha okay :) > > (In reply to comment #4) > > is this still a problem with 3.7? > > When I reported the bug was the only time I have experienced it. I'm > currently unable to produce this with hardened-sources-3.8.6. Okay thank you. This bug has become obsolete. There is no reason to return to the earlier version and figure out what was wrong. (In reply to comment #7) > Okay thank you. This bug has become obsolete. There is no reason to return > to the earlier version and figure out what was wrong. actually, there is ;). i took a look at the code and here's where it died: arch/x86/kernel/apic/io_apic.c:__setup_vector_irq 1252 »·······/* Mark the free vectors */ 1253 »·······for (vector = 0; vector < NR_VECTORS; ++vector) { 1254 »·······»·······irq = per_cpu(vector_irq, cpu)[vector]; 1255 »·······»·······if (irq < 0) 1256 »·······»·······»·······continue; 1257 1258 »·······»·······cfg = irq_cfg(irq); // returned NULL 1259 »·······»·······if (!cpumask_test_cpu(cpu, cfg->domain)) 1260 »·······»·······»·······per_cpu(vector_irq, cpu)[vector] = -1; 1261 »·······} the call to irq_cfg is a wrapper around irq_get_chip_data which can and seemingly does return NULL and its callers check for it elsewhere, but in this file it's not done consistently, even in 3.8.7. so i suggest that you tell the kernel devs about it as i think it's a problem in vanilla itself. i can add the obvious NULL checks but i don't know what the correct reaction to them is in each case. Thanks pipacs! (In reply to PaX Team from comment #8) > so i suggest that you > tell the kernel devs about it as i think it's a problem in vanilla itself. So did anyone tell the kernel devs or is this also already irrelevant at this point? |