d/numam-spdk
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

markdownlint: enable rule MD007

Browse Source 
MD007 - Unordered list indentation
Fixed all errors

Signed-off-by: wawryk <maciejx.wawryk@intel.com>
Change-Id: I051556bbd6bdaa94cec21bb072316bc693aea2ec
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/8998
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
This commit is contained in:
wawryk 2021-07-30 12:13:05 +02:00 committed by Tomasz Zawadzki
parent e96beb9374
commit fb9b29e732
2 changed files with 10 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
doc/usdt.md
View File

@ -177,28 +177,24 @@ print the final results of the maps.
itself as a probe parameter and let the script decide which fields it wants to access
(Note: these would need to be kept up-to-date with the C definitions of the struct - it is
not possible to include the header files in a bpftrace script)
- investigate using pahole to generate data structure definitions that can be included in
bpftrace scripts; this would allow us to pass the subsystem pointer itself as a probe
argument, and let the script decide which fields it wants to access; for example,
`pahole -E -C spdk_nvmf_subsystem build/bin/spdk_tgt` gets us close to what we need,
but there are some limiters:
- our structures have char arrays (not char pointers) for things like subnqn; large
arrays like these cannot currently be passed to bpftrace printf without generating
a stack space error (probe points are limited to 512 bytes of stack); we could
modify SPDK to have the char array for storage, and a char pointer that points to
that storage, the latter could easily then be used in bpftrace scripts
- our structures include fields with their enum types instead of int; bpftrace will
complain it does not know about the enum (pahole doesn't print out enum
descriptions); information on enums can be found in the applications .debug_info
section, but we would need something that can convert that into a file we can
include in a bpftrace script
- our structures have char arrays (not char pointers) for things like subnqn; large
arrays like these cannot currently be passed to bpftrace printf without generating
a stack space error (probe points are limited to 512 bytes of stack); we could
modify SPDK to have the char array for storage, and a char pointer that points to
that storage, the latter could easily then be used in bpftrace scripts
- our structures include fields with their enum types instead of int; bpftrace will
complain it does not know about the enum (pahole doesn't print out enum
descriptions); information on enums can be found in the applications .debug_info
section, but we would need something that can convert that into a file we can
include in a bpftrace script
- Note that bpftrace prints are not always printed in exact chronological order; this can
be seen especially with spdk_for_each_channel iterations, where we execute trace points
on multiple threads in a very short period of time, and those may not get printed to the
console in exact order; this is why the nvmf.bt script prints out a msec.nsec timestamp
so that the user can understand the ordering and even pipe through sort if desired
- flesh out more DTrace probes in the nvmf code

1
mdl_rules.rb
View File

@ -1,7 +1,6 @@
all
exclude_rule 'MD003'
exclude_rule 'MD004'
exclude_rule 'MD007'
exclude_rule 'MD009'
exclude_rule 'MD010'
exclude_rule 'MD011'