Fix UBSAN's cry afoul

[KungFuJesus]

Technically, we weren't actually doing this the way C wants us to,
legally. The zmemcpy's turn into NOPs for pretty much all > 0
optimization levels and this gets us defined behavior with the
sanitizer, putting the optimized load by arbitrary alignment into the
compiler's hands instead of ours.

[codecov]

Codecov Report

Merging #1194 (4184034) into develop (276a3ec) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff            @@
##           develop    #1194   +/-   ##
========================================
  Coverage    87.05%   87.06%           
========================================
  Files          113      113           
  Lines         9686     9692    +6     
  Branches      2462     2462           
========================================
+ Hits          8432     8438    +6     
- Misses        1008     1009    +1     
+ Partials       246      245    -1     

Flag	Coverage Δ
macos_clang	26.66% <ø> (ø)
macos_gcc	70.26% <50.00%> (-0.02%)	⬇️
ubuntu_clang	85.06% <50.00%> (-0.04%)	⬇️
ubuntu_clang_debug	84.24% <50.00%> (-0.03%)	⬇️
ubuntu_clang_inflate_allow_invalid_dist	84.92% <50.00%> (+0.61%)	⬆️
ubuntu_clang_inflate_strict	85.00% <50.00%> (-0.03%)	⬇️
ubuntu_clang_mmap	85.12% <50.00%> (+0.60%)	⬆️
ubuntu_clang_pigz	39.74% <50.00%> (+0.01%)	⬆️
ubuntu_clang_pigz_no_optim	40.22% <ø> (ø)
ubuntu_clang_pigz_no_threads	39.33% <50.00%> (+0.01%)	⬆️
ubuntu_clang_reduced_mem	85.25% <50.00%> (-0.04%)	⬇️
ubuntu_gcc	72.27% <50.00%> (-0.03%)	⬇️
ubuntu_gcc_aarch64	71.87% <ø> (ø)
ubuntu_gcc_aarch64_compat_no_opt	69.14% <ø> (ø)
ubuntu_gcc_aarch64_no_acle	70.05% <ø> (ø)
ubuntu_gcc_aarch64_no_neon	70.01% <ø> (ø)
ubuntu_gcc_armhf	71.86% <ø> (ø)
ubuntu_gcc_armhf_compat_no_opt	69.01% <ø> (ø)
ubuntu_gcc_armhf_no_acle	71.87% <ø> (ø)
ubuntu_gcc_armhf_no_neon	71.79% <ø> (ø)
ubuntu_gcc_armsf	71.86% <ø> (ø)
ubuntu_gcc_armsf_compat_no_opt	69.01% <ø> (ø)
ubuntu_gcc_benchmark	73.83% <50.00%> (-0.03%)	⬇️
ubuntu_gcc_compat_no_opt	71.14% <ø> (ø)
ubuntu_gcc_compat_sprefix	73.18% <50.00%> (-0.03%)	⬇️
ubuntu_gcc_mingw_i686	0.00% <0.00%> (ø)
ubuntu_gcc_mingw_x86_64	0.00% <0.00%> (ø)
ubuntu_gcc_no_avx2	70.93% <100.00%> (+0.01%)	⬆️
ubuntu_gcc_no_ctz	73.05% <ø> (ø)
ubuntu_gcc_no_ctzll	72.82% <ø> (ø)
ubuntu_gcc_no_pclmulqdq	69.74% <50.00%> (-0.02%)	⬇️
ubuntu_gcc_no_sse2	70.69% <100.00%> (+0.01%)	⬆️
ubuntu_gcc_no_sse4	70.57% <50.00%> (-0.02%)	⬇️
ubuntu_gcc_o3	73.23% <50.00%> (ø)
ubuntu_gcc_osb	∅ <ø> (∅)
ubuntu_gcc_pigz	37.70% <50.00%> (-0.01%)	⬇️
ubuntu_gcc_pigz_aarch64	38.28% <ø> (ø)
ubuntu_gcc_ppc	72.62% <ø> (ø)
ubuntu_gcc_ppc64	73.40% <ø> (ø)
ubuntu_gcc_ppc64le	72.55% <ø> (ø)
ubuntu_gcc_ppc_no_power8	73.52% <ø> (ø)
ubuntu_gcc_s390x	73.54% <ø> (ø)
ubuntu_gcc_s390x_dfltcc	72.32% <ø> (ø)
ubuntu_gcc_s390x_dfltcc_compat	69.91% <ø> (ø)
ubuntu_gcc_s390x_no_crc32	73.29% <ø> (ø)
ubuntu_gcc_sparc64	73.41% <ø> (ø)
ubuntu_gcc_sprefix	73.00% <50.00%> (-0.03%)	⬇️
win64_gcc	72.17% <50.00%> (-0.03%)	⬇️
win64_gcc_compat_no_opt	71.84% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
arch/x86/chunkset_avx.c	100.00% <100.00%> (ø)
arch/x86/chunkset_sse2.c	44.44% <100.00%> (+11.11%)	⬆️
gzlib.c	69.33% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 276a3ec...4184034. Read the comment docs.

[nmoinvaz]

Only suggestion is to use a different variable name than tmp. At work we try to not name any variables temp.

Maybe call it set1 or broadcast or repeat..

Other than that it looks good.

[KungFuJesus]

I was just going on the precedent set by the power8 and aarch64 ones that did the same thing.

[nmoinvaz]

Ah if they do the same then it can wait for another commit if we want to clean it all up later on. Anyways, I have already reviewed it and marked it approved. Good find.

[Dead2]

LGTM