Re: linux-next: Tree for Apr 9 (x86 boot problem)

[Randy Dunlap]

On 4/12/21 10:01 AM, Mike Rapoport wrote:
> On Mon, Apr 12, 2021 at 08:49:49AM -0700, Randy Dunlap wrote:
>> On 4/11/21 11:14 PM, Mike Rapoport wrote:
>>> Hi Randy,
>>>
>>> On Sun, Apr 11, 2021 at 07:41:37PM -0700, Randy Dunlap wrote:
>>>> On 4/9/21 4:51 AM, Stephen Rothwell wrote:
>>>>> Hi all,
>>>>>
>>>>> Changes since 20210408:
>>>>>
>>>>
>>>> Hi,
>>>>
>>>> I cannot boot linux-next 20210408 nor 20210409 on an antique
>>>> x86_64 laptop (Toshiba Portege).
>>>>
>>>> After many failed tests, I finally resorted to git bisect,
>>>> which led me to:
>>>>
>>>> # bad: [4c674481dcf9974834b96622fa4b079c176f36f9] x86/setup: Merge several reservations of start of memory
>>>> git bisect bad 4c674481dcf9974834b96622fa4b079c176f36f9
>>>>
>>>>
>>>> I reverted both of these patches and the laptop boots successfully:
>>>>
>>>> commit a799c2bd29d19c565f37fa038b31a0a1d44d0e4d
>>>> Author: Mike Rapoport <rppt@xxxxxxxxxx>
>>>> Date:   Tue Mar 2 12:04:05 2021 +0200
>>>>
>>>>     x86/setup: Consolidate early memory reservations
>>>>
>>>> &&
>>>>
>>>> commit 4c674481dcf9974834b96622fa4b079c176f36f9
>>>> Author: Mike Rapoport <rppt@xxxxxxxxxx>
>>>> Date:   Tue Mar 2 12:04:06 2021 +0200
>>>>
>>>>     x86/setup: Merge several reservations of start of memory
>>>>
>>>>
>>>> There is no (zero, nil) console display when I try to boot
>>>> next 0408 or 0409. I connected a USB serial debug cable and
>>>> booted with earlyprintk=dbgp,keep and still got nothing.
>>>>
>>>> The attached boot log is linux-next 20210409 minus the 2 patches
>>>> listed above.
>>>>
>>>> Mike- what data would you like to see?
>>>
>>> Huh, with no console this would be fun :)
>>> For now the only idea I have is to "bisect" the changes and move
>>> reservations one by one back to their original place until the system boots
>>> again. 
>>>
>>> I'd start with trim_snb_memory() since it's surely needed on your laptop
>>> and quite likely it is a NOP on other systems.
>>>
>>> diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
>>> index 776fc9b3fafe..dfca9d6b1aa6 100644
>>> --- a/arch/x86/kernel/setup.c
>>> +++ b/arch/x86/kernel/setup.c
>>> @@ -746,8 +746,6 @@ static void __init early_reserve_memory(void)
>>>  
>>>  	reserve_ibft_region();
>>>  	reserve_bios_regions();
>>> -
>>> -	trim_snb_memory();
>>>  }
>>>  
>>>  /*
>>> @@ -1081,6 +1079,8 @@ void __init setup_arch(char **cmdline_p)
>>>  
>>>  	reserve_real_mode();
>>>  
>>> +	trim_snb_memory();
>>> +
>>>  	init_mem_mapping();
>>>  
>>>  	idt_setup_early_pf();
>>>  
>>>> -- 
> 
> Hi Randy,
>  
>> Hi Mike,
>> That works fine.
>> Can you provide another/next step?
>  
> I thought about adding some prints to see what's causing the hang, the
> reservations or their absence. Can you replace the debug patch with this
> one:
> 
> diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
> index 776fc9b3fafe..a10ac252dbcc 100644
> --- a/arch/x86/kernel/setup.c
> +++ b/arch/x86/kernel/setup.c
> @@ -600,10 +600,13 @@ static bool __init snb_gfx_workaround_needed(void)
>  		return false;
>  
>  	vendor = read_pci_config_16(0, 2, 0, PCI_VENDOR_ID);
> +	devid = read_pci_config_16(0, 2, 0, PCI_DEVICE_ID);
> +
> +	pr_info("%s: vendor: %x, device: %x\n", __func__, vendor, device);

s/device)/devid)/

> +
>  	if (vendor != 0x8086)
>  		return false;
>  
> -	devid = read_pci_config_16(0, 2, 0, PCI_DEVICE_ID);
>  	for (i = 0; i < ARRAY_SIZE(snb_ids); i++)
>  		if (devid == snb_ids[i])
>  			return true;
> @@ -747,7 +750,7 @@ static void __init early_reserve_memory(void)
>  	reserve_ibft_region();
>  	reserve_bios_regions();
>  
> -	trim_snb_memory();
> +	pr_info("%s: snb_gfx: %d\n", __func__, snb_gfx_workaround_needed());
>  }
>  
>  /*
> @@ -1081,6 +1084,8 @@ void __init setup_arch(char **cmdline_p)
>  
>  	reserve_real_mode();
>  
> +	trim_snb_memory();
> +
>  	init_mem_mapping();
>  
>  	idt_setup_early_pf();

That prints:

[    0.000000] snb_gfx_workaround_needed: vendor: 8086, device: 126
[    0.000000] early_reserve_memory: snb_gfx: 1
...
[    0.014061] snb_gfx_workaround_needed: vendor: 8086, device: 126
[    0.014064] reserving inaccessible SNB gfx pages


The full boot log is attached.


-- 
~Randy

Attachment: boottest002.log.gz
Description: application/gzip