linux-kernel - Re: [PATCH V1 10/19] perf tools: Create maps for x86 PTI entry trampolines

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <29315e2d-a7cb-7c4e-5933-ce7b531b130b@intel.com>
Date:   Wed, 16 May 2018 14:37:08 +0300
From:   Adrian Hunter <adrian.hunter@...el.com>
To:     Jiri Olsa <jolsa@...hat.com>
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Andy Lutomirski <luto@...nel.org>,
        "H. Peter Anvin" <hpa@...or.com>, Andi Kleen <ak@...ux.intel.com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Joerg Roedel <joro@...tes.org>, linux-kernel@...r.kernel.org,
        x86@...nel.org
Subject: Re: [PATCH V1 10/19] perf tools: Create maps for x86 PTI entry
 trampolines

On 16/05/18 14:06, Jiri Olsa wrote:
> On Tue, May 15, 2018 at 03:43:24PM +0300, Adrian Hunter wrote:
> 
> SNIP
> 
>> +int machine__create_extra_kernel_map(struct machine *machine,
>> +				     struct dso *kernel,
>> +				     struct extra_kernel_map *xm)
>>  {
>>  	struct kmap *kmap;
>>  	struct map *map;
>> @@ -923,9 +915,33 @@ static u64 find_entry_trampoline(struct dso *dso)
>>  int machine__map_x86_64_entry_trampolines(struct machine *machine,
>>  					  struct dso *kernel)
>>  {
>> -	u64 pgoff = find_entry_trampoline(kernel);
>> +	struct map_groups *kmaps = &machine->kmaps;
>> +	struct maps *maps = &kmaps->maps;
>>  	int nr_cpus_avail = 0, cpu;
>> +	bool found = false;
>> +	struct map *map;
>> +	u64 pgoff;
>> +
>> +	/*
>> +	 * In the vmlinux case, pgoff is a virtual address which must now be
>> +	 * mapped to a vmlinux offset.
>> +	 */
>> +	for (map = maps__first(maps); map; map = map__next(map)) {
>> +		struct kmap *kmap = __map__kmap(map);
>> +		struct map *dest_map;
>> +
>> +		if (!kmap || !is_entry_trampoline(kmap->name))
>> +			continue;
>> +
>> +		dest_map = map_groups__find(kmaps, map->pgoff);
>> +		if (dest_map != map)
>> +			map->pgoff = dest_map->map_ip(dest_map, map->pgoff);
>> +		found = true;
> 
> hum, could you please explain/comment on above? the purpose of pgoff for these maps
> 
> all the trampoline maps seems to have the same pgoff at the end..
> 
> 0 0x158 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000006000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x1b0 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000032000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x208 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000005e000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x260 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000008a000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x2b8 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00000b6000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x310 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00000e2000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x368 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000010e000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x3c0 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000013a000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x418 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000166000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x470 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000192000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x4c8 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00001be000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x520 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00001ea000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x578 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000216000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x5d0 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000242000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x628 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000026e000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x680 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000029a000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x6d8 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00002c6000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x730 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00002f2000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x788 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000031e000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x7e0 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe000034a000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x838 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe0000376000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x890 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00003a2000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x8e8 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00003ce000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline
> 0 0x940 [0x58]: PERF_RECORD_MMAP -1/0: [0xfffffe00003fa000(0x1000) @ 0xffffffff8fc02000]: x __entry_SYSCALL_64_trampoline

The trampolines all map to the same page.  Since that page is also in the
kernel text, its virtual address (0xffffffff8fc02000 in your case) is also
recorded.  Then to find the offset in vmlinux,
machine__map_x86_64_entry_trampolines() can map that virtual address.